MA Thesis - O
MA Thesis - O MA Thesis - O
The Albanian Middle Construction MA Thesis Linguistics Department Radboud University Nijmegen September, 2011 Borana Lushaj s4081625 Supervisor: Prof. Helen de Hoop
- Page 3 and 4: To Shannon Woodcock La Trobe Univer
- Page 6: Table of Contents Chapter 1: Introd
- Page 9 and 10: presented below, the syntactic mani
- Page 11 and 12: a middle interpretation in their th
- Page 13 and 14: example, the following sentence can
- Page 15 and 16: grammatical subject of the phrase i
- Page 17 and 18: constructions. He supports an even
- Page 19 and 20: through interpretive mechanisms tha
- Page 21 and 22: logical direct object in its subjec
- Page 23 and 24: languages, the middle interpretatio
- Page 25 and 26: distinguishes between factors that
- Page 27 and 28: covered in previous treatments and,
- Page 29 and 30: picture of the construction as it i
- Page 31 and 32: In a context where the pragmatic fe
- Page 33 and 34: I suggest that the middle interpret
- Page 35 and 36: connections in terms of how these c
- Page 37 and 38: In determining whether the generic
- Page 39 and 40: iv. Generic Middle Construction: ag
- Page 41 and 42: formally transitive verb, the agent
- Page 43 and 44: Chapter 4 Generic Middle Constructi
- Page 45 and 46: Several difficulties were encounter
- Page 47 and 48: A while back, Ben Blushi, in descri
- Page 49 and 50: . Në fshatin tim... zërat in vill
- Page 51 and 52: It is understandable/It is understo
The Albanian Middle Construction<br />
<strong>MA</strong> <strong>Thesis</strong><br />
Linguistics Department<br />
Radboud University Nijmegen<br />
September, 2011<br />
Borana Lushaj<br />
s4081625<br />
Supervisor: Prof. Helen de Hoop
To Shannon Woodcock<br />
La Trobe University, Australia<br />
3
Acknowledgements<br />
The <strong>MA</strong> program at the Radboud Universiteit in Nijmegen proved to be a superb<br />
learning experience. I would like to thank my wonderful supervisor, Prof. Helen<br />
de Hoop, who has shown endless patience with me during the completion of this<br />
thesis. She has gracefully given much appreciated advice, and at the same time<br />
has given me much freedom in my roaming around and through the issue of<br />
middle generic constructions in Albanian. I would also like to thank my second<br />
reader for this thesis, Kees de Schepper, who provided insightful comments and<br />
showed interest in this work.<br />
I am grateful to the entire Linguistics Department at RU, including Dr. Ad Foolen<br />
for his readiness to give useful advice, Sander Lestrade, who helped me in the<br />
very beginning of this thesis with analysis the corpus, as well as Joshua Birchall,<br />
who unknowingly gave me some very useful advice, which helped me clarify my<br />
vision for this thesis.<br />
Thanks are due also to my internship advisor, Dr. Leah Roberts, who went out of<br />
her way to enable me to carry out my internship at the Max Planck Institute of<br />
Psycholinguistics in Nijmegen.<br />
Finally, I would like to thank my friends Katrin, Bram and Stefanie for making<br />
my stay in Nijmegen as pleasant and rewarding as it was as well as my family for<br />
their wonderful support.<br />
4
Table of Contents<br />
Chapter 1: Introduction.................................................................................................................8<br />
1.1 Introducing the middle generic construction...........................................................8<br />
1.2. Structure of the thesis.................................................................................................... 10<br />
1.3. Why are middle generic constructions interesting? .......................................... 11<br />
Chapter 2: Overview of Past Studies ..................................................................................... 16<br />
2.1. Early treatments of middle generic construction in English .......................... 16<br />
2.1.1. Summary..................................................................................................................... 19<br />
2.2. Generative accounts of the middle generic construction................................. 20<br />
2.2.1. Discussion ................................................................................................................. 26<br />
Chapter 3: The Generic Agent and Property Reporting in Albanian Middles........ 30<br />
3.1. The middle interpretation............................................................................................ 30<br />
3.2. Interpreting the generic agent in Albanian............................................................ 33<br />
Chapter 4: Generic Middle Constructions in Albanian ................................................... 43<br />
4.1. The corpus.......................................................................................................................... 43<br />
4.2. Methodology...................................................................................................................... 44<br />
4.3 Results.................................................................................................................................. 47<br />
4.4. Analysis and discussion................................................................................................. 55<br />
4.5. Conclusions ........................................................................................................................ 58<br />
Chapter 5: General Conclusions and Future Research ................................................... 60<br />
Bibliography.................................................................................................................................... 62<br />
Appendix: Generic middle constructions from corpus .................................................. 66<br />
6
Chapter 1<br />
Introduction<br />
1.1 Introducing the middle generic construction<br />
The present thesis investigates the middle generic construction in the Albanian<br />
language. Albanian is part of the Indo-European family and is spoken by about 7<br />
million people mainly in Albania, Kosovo, Western Macedonia, Southern<br />
Montenegro as well as in other parts of the world owing to its relatively large<br />
Diaspora (Lewis, 2009). A first example of a middle generic construction in<br />
English and its Albanian translation are given below:<br />
(1) This book reads nicely<br />
(2) Ky libër lexo-het kollaj<br />
this book read-NACT.3SG easily<br />
this book reads easily<br />
The middle generic construction in Albanian, just like the same construction in<br />
many of the IE languages, is parasitic on non-active morphology (Condoravdi<br />
1989; Lekakou, 2005), and its interpretation has been defined as denoting a<br />
property of a grammatical subject 1 . This interpretation is not very specific or<br />
conclusive given that many types of sentences have an interpretation which<br />
expresses a property of the grammatical subject, but not all of them are middle<br />
generic constructions. As it has been noted in various studies, some of which are<br />
1 Although in Albanian and Greek the middle generic construction relies on non-active morphology, its<br />
interpretation should not be confused with middle verbs, which denote actions with ambiguous/vague<br />
transitivity and which also make use of the same non-active verbal marking. These include, but are not<br />
limited to, verbs of grooming or body movement.<br />
8
presented below, the syntactic manifestation of a middle generic interpretation<br />
is also diverse cross-linguistically. Therefore relying on a combination of<br />
syntactic and semantic properties of the middle generic construction does not<br />
encompass the variety with which it is manifested in natural languages. The goal<br />
of this thesis is to study the middle generic construction in Albanian as is found<br />
in natural language, in the hope that surveying its realization in this language<br />
can contribute to a better understanding of its interpretation. For this purpose, a<br />
corpus of about 600 000 words 2 has been investigated, consisting of written<br />
texts of various genres, from fiction to political essays and includes diverse<br />
Albanian contemporary authors, as well as two translations into Albanian of<br />
English and Portuguese books. The focus of the thesis will be a specific type of<br />
middle generic construction, which, for lack of a better term, I will term a bare 3<br />
middle generic construction, as exemplified in (3) 4 :<br />
(3) Molla ha-het<br />
apple.NOM.SG<br />
the apple is edible<br />
eat-NACT.3SG<br />
The reason for this choice is that the addition of a manner adverbial makes most<br />
middle constructions in Albanian acceptable: in most types of sentences which<br />
have a third person non-active imperfective verb that agrees with the<br />
grammatical subject (i.e., the grammatical structure of the middle construction<br />
in Albanian) the middle generic construction succeeds in producing the middle<br />
interpretation. However, only a restricted number of verbs can be in this<br />
construction without an adverbial modifier. Therefore, I found it more<br />
2 The corpus, which was initially estimated to be about 1,200,000 words, was diminished in size for the<br />
final analysis, due to time constraints.<br />
3 I thank Kees de Schepper for suggesting this term<br />
4 Lekakou (2002, 2005) considers (3) a middle construction as is found in Greek<br />
9
interesting to explore constructions such as the one in (3) and investigate what<br />
makes the middle interpretation possible in these constructions, which seem to<br />
have less 'variables' to account for.<br />
1.2. Structure of the thesis<br />
In the following section of this chapter, I give an overview of the importance of<br />
the middle construction for the exploration of the relationship between syntax<br />
and semantics. In chapter 2 I provide an overview of some of the most important<br />
studies regarding the middle construction. While for languages like English,<br />
Dutch and German as well as French, the middle has been analysed as its own<br />
type of construction, linguists focusing on Albanian and Greek have analyzed it<br />
as a type of impersonal passive, based on the apparent uniformity with passives<br />
in terms of verbal morphology and the lack of a specific agent. In Chapter 3 I<br />
suggest a working definition of the middle interpretation for Albanian based on<br />
previous treatments as well as on the specifics of the data. I propose that the<br />
core characteristic that needs to be realized in order to obtain a middle<br />
interpretation is the modality-of-ability reading in the presence of a generic<br />
agent-role. In Chapter 4 I present the data, the methodology used for studying it,<br />
as well as some discussion on the problems with the methodology and with the<br />
corpus. In the section 3 of Chapter 4 and onwards I present the findings of the<br />
investigation and analyze its implications. Based on our data, it turns out that in<br />
Albanian verbs which do not have a [+human] agent in their argument structure<br />
cannot produce a middle interpretation. Moreover, it seems that mainly verbs of<br />
cognition and perception can produce an ability reading in a non-modified (i.e.,<br />
bare) middle generic construction 5 . At a finer level of detail, even within the<br />
group of cognition and perception verbs, there are verbs which do not allow for<br />
5 Modifications include negation, adverbials as well as contrastive intonation<br />
10
a middle interpretation in their third person non-active imperfective forms. This<br />
data suggests that the ability reading which I equate with the middle<br />
interpretation in non-modified middles tends to be realized for those verbs<br />
which imply a process in their meaning. In general, the structural complexity of<br />
sentences where verbs can produce this modality of ability reading in Albanian<br />
suggests that this reading is due to the semantic properties of the verb, and<br />
grammatical and discourse changes that occur when the verb is used in the nonactive<br />
form. Chapter 5 concludes the project, where I present the general<br />
conclusions of this work as well as future research avenues on the middle<br />
construction.<br />
1.3. Why are middle generic constructions interesting?<br />
Below I present examples of one type of middle construction in English, Dutch,<br />
French, Greek, Italian, and Albanian, respectively (Lekakou, 2005; e. and f. are<br />
translations of the above):<br />
(4)<br />
a. This book/A large print book reads easily<br />
b. Dit boek leest gemakkelijk<br />
this book read.3SG easily<br />
c. Ce livre se lit facilement<br />
this book REFL reads easily<br />
d. Afto to vivlio diavazete efkola<br />
this the book read.PASS.3SG easily<br />
e. Questo libro si legge facilmente<br />
this book REFL read easily<br />
f. Ky libër lexo-het kollaj<br />
this book read-NACT.3SG easily<br />
11
The commonly accepted interpretation of the sentence above is that it expresses<br />
a generalized assertion about a property of the grammatical subject (Ackema &<br />
Schoorlemmer 2005). The subject of a middle construction, i.e., the entity that<br />
the generalized assertion is about, can be a specific or a ‘kind’ of object, as in<br />
(4a), a human entity as in (5), or a general event as in (6):<br />
(5) John scares easily<br />
(6) Graduating from a prestigious university reads nicely on a CV<br />
Even though the sentences above are grammatically acceptable for native<br />
speakers, my impromptu investigations for English suggest that they sound<br />
slightly odd at first, if the speaker is new to this type of sentence. Clearly,<br />
linguists did not invent the construction, but it can be safely said that they might<br />
be the only group of individuals worldwide who are completely at ease with it,<br />
and are keen on understanding it further. And there is a reason why linguists are<br />
interested in this construction: it is a testing ground for theories of syntax and<br />
semantics and their interaction (Fagan, 1992) because its interpretation seems<br />
to be uniform but its syntactic manifestation is varied in different languages.<br />
From the syntactic point of view the middle construction is interesting, because,<br />
in some languages like English and Dutch, it manifests both properties of active<br />
and passive sentences (Lekakou, 2002). Firstly, the grammatical subject of the<br />
middle construction is, in the most popular examples, the logical object of the<br />
verb. At the same time, in languages like English and Dutch the verb stays in the<br />
active form, while in Albanian, Greek, Italian and French the verb is in its nonactive<br />
form. However, in Albanian and Greek we can have middle constructions<br />
with the verb both in the active and non-active form as pointed out by<br />
Condoravdi (1989) for Greek, where some verbs can form a middle construction<br />
in both, some only in the active form and some only in the non-active form. For<br />
12
example, the following sentence can be interpreted as a middle construction:<br />
(7) ky laps shkruan (mirë)<br />
this pencil write.3SG (well)<br />
this pencil writes (well) (i.e., it is comfortable to write with it)<br />
At more theory-internal levels, there is the question whether the construction is<br />
the result of a movement operation in the syntax, as is assumed by<br />
transformational accounts (Keyser & Roeper, 1984) or whether it involves the<br />
projection of pre-syntactic ‘features’ of the elements of the middle construction<br />
into syntax proper (Ackema & Schoorlemmer 1994), which is a topic we will<br />
cover more in depth in Chapter 2.<br />
From the semantic point of view, the middle construction provokes interest at<br />
various levels. In the field of lexical semantics it is of interest to understand<br />
whether a verb’s lexical conceptual structure allows for the middle<br />
interpretation to arise (Stalmaszczyk, 1993). A lexical conceptual structure<br />
captures the meaning components of a verb which are responsible for its<br />
syntactic behaviour (Levin & Rappaport-Hovav, 2008), i.e., what types of<br />
arguments can verbs take and in what types of constructions they can appear.<br />
For example, a verb such as ‘break’ can form an anticausative, i.e., a construction<br />
similar to the passive but where the agent is completely removed from the<br />
meaning (Haspelmath & Müller-Bardey, 2004). In a sentence like 'the glass<br />
broke', the agent is not only removed syntactically but it is also not necessarily<br />
implied in the interpretation of the sentence because the lexical conceptual<br />
structure of the verb ‘break’ contains the resulting state of being broken but<br />
does not specify any agent causing the change of state of the glass, as is<br />
illustrated below from Levin & Rappaport-Hovav (2008:4)<br />
13
(8) break: y come to be BROKEN (Guerssel et al. 1985: 54, ex. (19)<br />
With regards to the middle generic construction, the question is whether we can<br />
stipulate that they can be formed only with a specific class of verbs that have a<br />
specific lexical conceptual structure. From another viewpoint, middle generic<br />
constructions are one kind of generic sentence (Carlson & Pelletier, eds., 1995)<br />
and therefore it is plausible that they be also accounted for in terms of reference<br />
to 'kinds' of entities and 'properties' of entities.<br />
The middle interpretation in fact is still an open issue: what does the middle<br />
interpretation tell us? What state of affairs does it evoke? One view assumes<br />
that, given an entity or a state of affairs, what the middle interpretation recounts<br />
is the ability of a generic/potential human agent to perform the action described<br />
by the verb; this action in turn is somehow related to or affects the grammatical<br />
subject (Fagan 1992). Another view discusses the middle interpretation as<br />
conveying a generalized statement about events, and there is no need for an<br />
implied generic/potential agent (Condoravdi, 1989, Rapoport 1999a). As an<br />
illustration of this conflict, consider the following sentence:<br />
(9) Alloy blackens easily<br />
This is clearly a natural process and there is no human agent involved in its<br />
realization. If (9) is considered to have a middle interpretation, then the generic<br />
human agent who would perform the action is lacking in this instance. In<br />
addition, the role of the grammatical subject/topic of a middle construction is<br />
also in question. In various treatments (Keyser & Roeper 1984 and many others)<br />
the grammatical subject is described as the logical object of the verb. However,<br />
there are different types of middles in other languages which do not abide by<br />
this specific criterion. For example in Dutch there are middles where the<br />
14
grammatical subject of the phrase is not an underlying object. Consider the<br />
following sentence from Hoekstra & Roberts (1993:197):<br />
(10) Die stoel zit lekker<br />
the chair sits nicely<br />
the chair sits nicely (this is a comfortable chair)<br />
In addition, as it was pointed out in the introduction, in Greek and Albanian,<br />
middle generic constructions are well-formed without the need of an adverbial<br />
(Lekakou 2002, 2005):<br />
(11) Ky ujë pi-het<br />
this water drink-NACT.3SG<br />
this water is potable 6<br />
Finally, the middle construction enables a modality of ability interpretation<br />
(Ackema & Schoorlemmer, 2005; Lekakou, 2005), which has seldom been<br />
analysed in depth on its own. The modality of ability, in my understanding, can<br />
be termed as the relationship between the generic agent that is implied in<br />
middles and the predication.<br />
In conclusion, the middle construction presents many interesting aspects which<br />
require an explanation of how they come about. The different viewpoints in<br />
previous treatments are not mutually exclusive, nor are the answers convergent<br />
to a satisfactory degree from a cross-linguistic perspective. Although the<br />
construction is well studied, the topic is hardly exhausted and it provides scope<br />
for deeper investigations into its properties.<br />
6 This phrase also has other interpretations but these will be indicated below as here they would be<br />
outside the scope of the chapter<br />
15
Chapter 2<br />
Overview of Past Studies<br />
2.1. Early treatments of middle generic construction in English<br />
The middle construction seems to have been first made prominent as a special<br />
case of structural arrangement in the English language quite early in 20 th<br />
century linguistics. There was a time when it was not yet called a “middle<br />
construction”. Jespersen (1927) talks about “activo-passive” use of verbs,<br />
bringing the following examples to the table 7 : my plays won’t act; my poesy<br />
won’t sell; I am at a sentence that will not write; this filthy fluff will never brush<br />
off my bags; the figures made her cry…they would not add up; the coat had been<br />
tighter, ‘t would never have hooked nor buttoned; can compare with thee for the<br />
rare effects of magic; the meat cooks all the better if you cook it slow; alabaster<br />
cuts very smooth and easy; constitutional ill-health [does not] necessarily<br />
transmit to a child; the ham was not digesting very well; reaming swats (ale) that<br />
drank divinely; one of our French wither’d pears, it looks ill, it eats drily; if the<br />
cakes at tea eat short and crisp, they were made by Olivia; the story told well;<br />
the words would not form on her lips; we might put up two or three [big houses]<br />
and see how they let; that doesn’t listen so bad; four babies, none of whom<br />
photographed well; the old gentleman’s speech reads excellently; the dirt rubs<br />
off 8 .<br />
Jespersen’s analysis particularly hinges on the interpretation of these<br />
7 Providing these examples from the original authors seems useful for readers who still need to familiarise<br />
with the different manifestations of this construction in English.<br />
8 The examples are not kept in the alphabetical order as in Jespersen’s book, and some orthographic<br />
changes have been made to fit today’s orthographic rules of English.<br />
16
constructions. He supports an even older idea that the grammatical subject of a<br />
middle construction (or activo-passive construction in his terminology) is in fact<br />
the logical object of the verb, while the logical subject itself is not specified<br />
because it is indefinite (cf. Sweet, 1986). On the other hand, Jespersen suggests<br />
that even though the grammatical subject of a middle construction is the object<br />
of the predicate, it also serves as some kind of agent because the properties of<br />
this subject are what the construction describes; it is in virtue of something that<br />
is “characteristic” of the subject that the construction makes sense 9 . In his own<br />
words "when we say ‘his novels sell very well’ we think to some extent of the<br />
books as active themselves, as the cause of the extensive sale, while we do not<br />
think so much of the activity of the bookseller" (Jespersen, 1927: 350).<br />
Therefore, he places no restrictions on the verb that can be in this construction,<br />
but states that there are special circumstances (which we can interpret as<br />
conceptual and contextual) which apply to the well-formedness of the middle<br />
construction. A last interesting point from this early take on the middle<br />
construction is the proposal that Jespersen puts forward about the origin of the<br />
construction. He sees the origin of the middle in an extension of the use of<br />
gerundive forms, which do not distinguish between active or passive in their<br />
formation rules: ‘the book is selling well’ → ‘the book sells well’. In gerundive<br />
forms as the one in the example, the subject of the verb does not have to be a<br />
human agent but it can also be an object as well as many other types of entities.<br />
A second early treatment of the middle construction is Hatcher (1943)’s paper<br />
on the construction, which she terms a ‘derived intransitive’. In O’Grady’s (1980)<br />
summary of this paper, she is also quoted as supporting the view that the<br />
‘derived intransitive’ "is used to emphasize some property inherent in the<br />
ontological patient" (O’Grady, 1980, p. 62). In her paper, Hatcher differentiates<br />
9 We might think of this statement as a first hint at the ‘characterizing generic’ nature of middle<br />
construction (Carlson & Pelletier, eds., 1995).<br />
17
etween what today is termed anticausative constructions, such as ‘the door<br />
opens’ and middle constructions, where the former can occur without the<br />
presence of an animate agent but the latter can't. Moreover, she suggests that the<br />
construction does not describe "the activity of this agent but an attribute of the<br />
subject" (p. 9), or "the possibility of such activity" (Hatcher, 1943, p. 10). In other<br />
words, the predicate expresses “a way of realizing the potentialities of the<br />
object” (p.11). Here we find a first approximation of what in current literature is<br />
referred to as the generic agent in middle constructions, which Hatcher labels as<br />
a “qui que se soit” human (p.12). While Hatcher interprets the middle as<br />
describing a property of the grammatical subject, the realization of this<br />
interpretation seems to be crucially dependent on the present of the animate<br />
agent that can hypothetically perform the action. She brings examples from the<br />
advertisements of the time, explaining how this construction is becoming more<br />
frequent at the time due to its use in advertising campaigns: Couches convert<br />
easily into beds; Bed-lamps attach and adjust easily; Drawers pull out and trays<br />
lift out easily; Faucets turn on and off easily; The clock winds easily; Lingerie<br />
tubs quickly and irons easily; Garments pack and unpack neatly; Automobiles<br />
steer and park easily; Cream whips quickly; Paint applies evenly; Nail-polish<br />
removes easily; Bond paper erases neatly; Linoleum wipes off easily (p.12-13).<br />
Hatcher also notes that the grammatical subject of a middle construction can<br />
also be an animate, e.g. (1) I don’t scare easily; (2) I don’t surprise so easy<br />
(p.14). 10<br />
O’Grady (1980) is influenced by these early accounts and states that the derived<br />
intransitive construction (i.e., middle construction) should be accounted for in<br />
terms of interpretive constraints rather than pre-interpretive ones, since any<br />
‘production’ model would posit problems of over-generation, and it is only<br />
10 For a more detailed historical overview of the middle construction in English, see O'Grady (1980) and<br />
Stalmaszczyk (1993) from which this summary benefitted.<br />
18
through interpretive mechanisms that over-generation can be constrained. Such<br />
interpretive mechanisms involve actualization of the predication and facilitating<br />
such actualization. Hence, the derived intransitive in O'Grady's terms involves a<br />
situation which describes inherent properties of the grammatical subject and<br />
therefore the elements in this construction refer to the potential/hypothetical<br />
actualization of such property by an agent that is related to the grammatical<br />
subject in a manner that is coherent to the grammatical subject, its properties, as<br />
well as the agent him/herself. He also emphasizes that only verbs which imply<br />
the presence of an agent can occur in a derived intransitive construction, giving<br />
the example of ‘wear’ vs. ‘wear out’, where, he claims, the latter can’t be in such<br />
construction. Summarizing, he posits that the well-formedness conditions on<br />
derived intransitives “seem to result from the subtle interaction of the concept of<br />
actualization with the semantic properties associated with the referent of the<br />
grammatical subject” (p. 70).<br />
2.1.1. Summary<br />
Early accounts of the middle construction stress three main points. Firstly, the<br />
grammatical subject needs to have semantic coherence with the action described<br />
by the verb. This coherence can be thought of as one or a number of properties<br />
inherent in the meaning of the grammatical subject to which the middle<br />
construction applies. The 'responsibility' of the grammatical subject in the<br />
construction is crucial to its well-formedness. Secondly, especially stressed in<br />
Hatcher (1943), the middle construction’s well-formedness is contingent upon<br />
the existence of an implied generic agent who can actualize the event described<br />
by the clause. Thirdly, as stressed in O’Grady (1980), the middle construction<br />
should be accounted for more in terms of its interpretation than from its surface<br />
grammatical form, which varies across language, and which, even within a single<br />
language, needs to be semantically and contextually motivated.<br />
19
2.2. Generative accounts of the middle generic construction<br />
Two views are prominent in the middle construction analysis. One view offers an<br />
account of the middle at the syntax-semantics interface, where the properties of<br />
the middle can be explained by the semantic-syntactic properties of verbs and<br />
the grammatical subject in a middle construction. Indeed, the middle motivates<br />
different grammatical architectures of the interaction between different syntax<br />
and the lexicon and has proved a feast for the generative tradition, because the<br />
well-formedness of the construction seems to rely on the semantic properties of<br />
verbs and their arguments. 11<br />
Keyser and Roeper (1984) propose an account of middles that is similar to the<br />
passive formation in a generative transformational framework. They analyse<br />
middles vis-à-vis ergative verbs while accepting that both transitive and ergative<br />
verbs can be successfully used in a middle construction. Ergative verbs are those<br />
verbs which can suppress the causer of an action completely such as the verb<br />
‘break’ 12 . Thus, for them, the following are both middle constructions:<br />
(12) This door opens easily<br />
(13) This dress buttons easily<br />
They suggest that verbs which can be both in a middle and in an anticausative<br />
construction (such as 'open'), the so-called labile verbs, realize two argument<br />
structures, a transitive and an intransitive one. The anticausative construction is<br />
formed by the intransitive member of such pairs while the middle construction is<br />
formed by the transitive member. From this account, only transitive verbs in<br />
English can enter a middle construction, stipulating that the middle must have a<br />
11 Of course, the theory-internal issue of what syntax and semantics encompass is at the core of this debate.<br />
12 The terms 'ergative’ and ‘anticausative’ will be used interchangeably throughout the thesis.<br />
20
logical direct object in its subject position.<br />
Ackema & Schoorlemmer (1994) however, point out that middles in Dutch do<br />
not involve movement because they appear to be syntactically unergative.<br />
Starting from Perlmutter’s (1978) insight that in some languages unaccusatives<br />
in the perfect tense take the auxiliary ‘be’ while unergative verbs take the<br />
auxiliary 'have', they point out that in Dutch, middle generic construction with a<br />
perfect tense verb take the auxiliary ‘have’, therefore pattern with unergatives,<br />
as illustrated in the example below (Ackema & Schoorlemmer, p. 61):<br />
(14) Dit vlees heft/*is altijd gemakkelijk gesneden<br />
this meat has/is always easily cut<br />
Since their assumption is that auxiliary selection depends on whether the D-<br />
structure subject is the surface subject or not, this shows that in Dutch the<br />
surface subjects of the middle generic construction are also underlying subjects<br />
at D-structure. Ackema and Schoorlemmer also acknowledge the presence of the<br />
generic agent, which they explain by assuming the projection of an arbitrary<br />
agent role from a two-tiered Lexical Conceptual Structure (LCS) of verbs<br />
containing a thematic tier and an action tier. In this architecture, the arguments<br />
which are specified in the LCS are then projected onto Deep Structure and then<br />
to Surface Structure. In their description, the thematic tier involves information<br />
on the thematic roles of the predicate like theme, agent, goal, etc. Overall,<br />
representations at the thematic tier only include information on spatio-temporal<br />
relations between arguments, e.g., causality. On the other hand, the argument<br />
structure of a predicate is determined at the next level of representation, i.e., the<br />
Action Tier. Here is where the relationships between arguments are specified<br />
and no argument is arbitrarily external, but it is assigned as such by a specific<br />
argument structure which is dictated by the action tier. Ackema & Schoorlemmer<br />
21
attribute this to the whole frame of the middle, which contains a feature (+ext)<br />
that externalizes the argument with the highest-ranking thematic role that is<br />
projected onto D-structure. In their treatment of the middle formation the agent<br />
in an middle construction is not suppressed as in a passive or an anticausative,<br />
but it is only assigned the feature ‘arbitrary’ at the level of the action tier, and it<br />
is recoverable precisely due to its arbitrary nature, or from the linguistic context.<br />
Lekakou (2005) uses the Principles and Parameters framework<br />
(Chomsky & Lasnik, 1993) for her account of the construction. She compares the<br />
middle constructions in English and Dutch with those in French and Greek,<br />
pointing out the main differences in how the middle construction is formed in<br />
the two groups of languages. Most significantly, in Greek the presence of an<br />
adverb is not obligatory as in English she claims, due to the fact that Greek<br />
encodes genericity morphologically while English and Dutch license genericity<br />
through a generic operator that is not present in syntax. In Greek, genericity is<br />
parasitic on the imperfective aspect making middle generic constructions more<br />
prevalent in spoken language. For Lekakou, in English and Dutch as well as in<br />
Greek and French, middles include a generic agent which she terms as *ONE. The<br />
difference between the languages is that in English and Dutch the generic agent<br />
is not projected onto syntax contrary to Greek and French. Precisely this fact<br />
makes middle formation more flexible in the latter two languages.<br />
The second view on middles analyses the construction from a strictly semantic<br />
viewpoint, arguing that the middle is a notion, an interpretation which is realized<br />
differently in different languages (I provided examples of this in the<br />
introduction) and therefore understanding its nature is the more pressing issue.<br />
Valuable arguments are offered on both sides. Condoravdi (1989) proposes an<br />
approach to the analysis of the middle construction that is reminiscent of<br />
O’Grady (1980), by suggesting that while the manifestation of the middle<br />
construction may be parasitic on the morpho-syntactic properties of different<br />
22
languages, the middle interpretation as a notional category is the same in all<br />
languages. Following the treatment of generic sentences by various authors,<br />
including Heim (1982), Schubert & Pelletier (1989), Krifka (1987b), she applies<br />
a similar semantic analysis to middle sentences. Generic sentences are assumed<br />
to have a tripartite structure represented in their logical form consisting of a<br />
generic operator, a restrictive clause and a nuclear scope. In this analysis the<br />
following structure would be applied to the sentence (Condoravdi, 1989, p.18):<br />
(15) This bread cuts smoothly<br />
[e: bread (x), cut (e), Patient (e, x)] [smooth (e)]<br />
It seems then that the generalization in middles is over events in the nuclear<br />
scope (i.e., the adverb in the 'classical' middle construction), the latter also<br />
allowing for contextual and presupposed information to be integrated.<br />
Paraphrasing Ackema & Schoorlemmer's (2005) review of this work,<br />
Condoravdi's interpretation of (15) would be something along the lines of “in<br />
general, in events in which this bread is cut, the event is ‘smooth’”. This account<br />
explains the necessity for adverbials or some other kind of modification to the<br />
event in English middles, because the nuclear scope must be given content.<br />
Condoravdi does not support the line of research that tries to delimit a class of<br />
verbs that can be in a middle formation, due to its highly contextual nature.<br />
However, it must be noted that the analysis here does not take into account<br />
adverbless middles or middles without some other type of modification, like<br />
negation or contrastive focus, which are the elements that fill the nuclear scope<br />
of the generic middle sentence. At the same time, by providing an account of the<br />
middle interpretation only in terms of a generic statement about events, then the<br />
ability reading of the middle interpretation is lost. Indeed, it seems that by<br />
providing this account she includes as middles those sentences which she clearly<br />
23
exempts from the middle interpretation, such as sentences from Fellbaum &<br />
Zribi-Hertz (1987, in Condoravdi, 1989), who claim a middle interpretation for<br />
sentences which only can be said to have a generic reading, like:<br />
(16) Aristophane se traduit rarement<br />
Aristophanes REFL translates<br />
rarely<br />
dans les lycées<br />
in the high.schools<br />
Aristophanes is rarely translated in high schools.<br />
While I support the insight that the sentence above in not a middle construction,<br />
my suggestion is that it fails to produce a middle construction because it does<br />
not produce an ability reading. The French sentence in (16) is only a generic<br />
statement about the state of affairs regarding Aristophane’s books in the<br />
educational system of France. In my data-filtering process, where I have<br />
attempted to distinguish between ‘simply’ generic sentences with non-active<br />
verb forms and those which successfully incorporate a middle interpretation, I<br />
have applied precisely this distinction.<br />
Another extensive treatment of the middle construction is that of Fagan (1992),<br />
who focuses on English, German and French. She treats middle formation as a<br />
generalizing operation on object-like arguments, which is realized through the<br />
assignment of an arbitrary or indefinite implicit agent to the predication.<br />
Paraphrasing Ackema & Schoorlemmer (2005), Fagan's interpretation of the<br />
middle in (15) would be: “given this bread, anyone who will cut it will<br />
experience it as a smooth event”. For the realization of the middle interpretation,<br />
stress is placed on the accessibility of the implied generic agent. Fagan<br />
24
distinguishes between factors that enter into play in English and German<br />
middles vs. French middles, and summarizes her research on the constraints on<br />
middle formation. The implied agent in English and German is generally<br />
recovered owing to the fact that it is assigned an arbitrary feature in the logical<br />
form of the sentence. In French on the other hand, the implied agent can have<br />
both the features arbitrary and indefinite. The addition of the indefinite agent<br />
allows for the attested flexibility of the middle in French, where it is suggested<br />
that the interpretation of the construction is not necessarily “be able to be X-ed”<br />
but also just “be X-ed” but by an indefinite agent. This interpretation would<br />
consider a sentence like (16) under the definition of a middle generic<br />
construction. In fact, generic sentences that produce modalities like the<br />
prescriptive one and also eventive middles are subsumed under this hypothesis.<br />
With regards to restrictions on the formation of the middle, it is claimed that in<br />
German and English verbs need to be accomplishments or activities, in the<br />
lexical semantic framework provided by Vendler (1967), while in French, verbs<br />
that allow for middle formation must be agentive. Even though these two classes<br />
of verbs are not necessarily comparable given that they are used in different<br />
theoretical frameworks, we can for now agree that in English and German<br />
aspectual differences are more prominent in determining the grammatical<br />
properties of the verbs (i.e., in which construction they can appear) than in<br />
French, where agentivity plays a more prominent role. The reason why this is so<br />
it is unclear in Fagan’s treatment, just as it remains unclear to me why there<br />
should be two different interpretations of middles in English on the one hand,<br />
and in German and French on the other.<br />
25
2.2.1. Discussion<br />
The definition of the middle construction in the generative tradition<br />
encompasses different conditions on well-formedness proposed by different<br />
scholars. By well-formedness, we mean not only the grammatical constraints on<br />
the construction, such as the presence of a grammatical subject and contingent<br />
agreement of the predicate with the subject, but also conditions on the<br />
interpretation of the middle construction. The generally accepted properties of<br />
the middle generic construction are summarized by Ackema & Schoorlemmer<br />
(2005):<br />
(17) A: The subject of the sentence corresponds to the<br />
internal argument (the understood or notional object).<br />
B: The agent is demoted and receives an arbitrary<br />
interpretation.<br />
C: The interpretation of the sentence is non-episodic.<br />
Middles do not make reference to an actual event<br />
having taken place; rather, they report a property of<br />
the grammatical subject. The otherwise eventive verb<br />
becomes a derived stative and, more precisely, receives<br />
a generic modal interpretation.<br />
The first and second well-formedness conditions are self-explanatory while the<br />
third condition deserves a deeper analysis, namely, the aspect of reporting a<br />
property of the grammatical subject. How can this property reporting be<br />
characterized and how does it come about? In order to investigate this question,<br />
the following chapters of this thesis will focus on adverbless middle<br />
constructions in Albanian, a type of middle constructions which has seldom been<br />
26
covered in previous treatments and, of course, not with regards to Albanian.<br />
Ackema & Schoorlemmer (2005) indicate that these adverbless constructions<br />
are non-core middles as they put it, due to their lack of an adverbial, which<br />
construes a different interpretation. The difference in interpretation between a<br />
canonical middle and an adverbless middle is summarized in the table below<br />
(A&S 2005:154):<br />
(18)<br />
Type I middle (X V Adv):<br />
‘X has a property such that it is possible for<br />
anyone somehow to V X.’<br />
(e.g., this paper washes easily ‘this paper<br />
has a property such that it is possible for<br />
anyone to wash it with ease’)<br />
Type II middle without adverbial<br />
modification (X se-V):<br />
‘X has a property such that it is possible<br />
for X to be V’ed.’<br />
(e.g., ce papier se lave (65a) ‘this paper<br />
has a property such that it is possible for<br />
it to be washed’)<br />
Clearly, the difference between the two types of middles is supposed to be<br />
subtle. However, it seems that the difference proposed here distracts us from the<br />
main notion which these two middle types share, and which in fact is the key<br />
component of a middle interpretation: i.e., the property of the grammatical<br />
subject/topic is spelled out by the potential actions of a generic agent. In other<br />
words, in both types of middles something about the grammatical subject/topic<br />
is uttered, and this aboutness consists in how a generic agent can act with<br />
regards to the subject/topic, in virtue of the properties of said grammatical<br />
subject. Consider the following examples:<br />
27
(19) a. This dress buttons<br />
b. This dress buttons nicely/easily<br />
Intuitively, in 19(a) one interprets the dress as having a property: the property<br />
that someone (an implicit agent) buttons it when wearing it. In 19(b) we further<br />
expand the interpretation in (a) and understand that the dress, which can be<br />
buttoned, has the property of being pleasant to the eye because of the buttons /<br />
is easy to button. An attempt to paraphrase the interpretations in (19a) and<br />
(19b) is given in (20):<br />
(20) (a) This dress has buttons and that is how it is<br />
worn (while that other one zips up);<br />
(b) This dress has buttons (and that is how it is<br />
worn) but the buttoning has some other quality<br />
to it, e.g., it looks nice/it is an easy process.<br />
Thus it seems clear that (19a) is sufficiently able to convey a property of the<br />
grammatical subject, and therefore is a suitable candidate for being considered a<br />
middle generic construction. Further support for my interpretation of (19a)<br />
comes from the examples that were brought by Jespersen and Hatcher in chapter<br />
II. One can see that some of them do not have an adverbial modification, or even<br />
no modification at all, e.g., the dirt rubs off.<br />
In the rest of this thesis I will give examples of non-modified middle generic<br />
constructions without adverbial modification. However, it must be noted that in<br />
the data I have not excluded for the present purposes other types of modification<br />
which are often sufficient for the well-formedness of a middle construction.<br />
The reason for this modus operandi is purely pragmatic: I would like to provide<br />
broad resources in Albanian in this thesis, and especially provide a broader<br />
28
picture of the construction as it is found in text in order to be able to undertake<br />
adequate comparative analysis. In case the modification turns out to play a<br />
crucial role in the realization of the middle interpretation (i.e., if say negation is<br />
the crucial element that makes the middle interpretation possible), this will be<br />
highlighted in the analysis.<br />
29
Chapter 3<br />
The generic agent and property reporting in Albanian<br />
middles<br />
3.1. The middle interpretation<br />
In the previous chapters we have outlined a number of perspectives which focus<br />
on the semantic and syntactic properties of components of the middle generic<br />
construction. From this point onwards, we will shift our paradigm of<br />
investigation and will provide an interpretation of the middle generic<br />
construction from a functional perspective. The middle generic construction is<br />
reminiscent of impersonal constructions in that it constraints the interpretation<br />
of a generic agent. The difference between the two is that the middle generic<br />
construction also contains an element of modality. Indeed the middle generic<br />
construction in traditional Albanian grammars is described as an impersonal<br />
construction with a modality reading (Camaj 1984). An impersonal construction<br />
expresses a state of affairs where a generic agent is implied, as in 21(a), while a<br />
middle construction expresses the potentiality of a generic agent to perform the<br />
action, as in 21(b):<br />
(21) Këtu kërce-het<br />
here dance-NACT.3SG<br />
a. Here is where dancing usually happens/where people<br />
dance (generic)<br />
b. Here one can dance/is allowed to dance<br />
(middle generic)<br />
30
In a context where the pragmatic features of the discourse/situation provide for<br />
a modality reading, a reading such as 21(b) is obtained. For example, if one goes<br />
on a terrace that seems a bit shaky, one may ask in Albanian:<br />
(22) A kërce-het këtu?<br />
Q dance-NACT.3SG here<br />
Can one dance here? (i.e., does the terrace hold?/will the<br />
neighbors mind?)<br />
The same difference exists between an impersonal passive and a generic middle<br />
construction with a transitive verb. Let’s consider our first example of a bare<br />
generic middle, repeated here for convenience, where we add its various<br />
potential interpretations:<br />
(23) Molla ha-het<br />
apple.SG eat-NACT.3SG<br />
a. the apple is eaten<br />
b.. the apple is edible<br />
c. the apple should be eaten (it should not be thrown away/boiled<br />
for making jam)<br />
In this instance, we can have three interpretations: in 23(a) we understand a<br />
concrete event to have taken place by an unspecified agent; in 23(b) and 23(c)<br />
we understand that an event can be/ should be performed by an unspecified<br />
agent which is also generic. Kemmer (1993:147-8), in her cognitive-typological<br />
study of middle verbs suggests that the middle-generic construction is not part<br />
of the actual category of middle verbs, which were defined in footnote 1. She<br />
points out that, contrary to what its terminology suggests, in her perspective, the<br />
middle generic construction is closer in interpretation to a passive construction.<br />
31
Kemmer indicates that in middle generic constructions “the patient has initiator<br />
status…[because] the event is conceived as proceeding from the patient (i.e.,<br />
grammatical subject : B.L) by virtue of the inherent characteristics of that entity.<br />
In other words, the constituent about which something is predicated in the<br />
middle construction is the topic and the aboutness expressed is construed along<br />
its semantic and pragmatic properties. In my perspective, the aboutness that is<br />
reported in middle generic constructions as the one above can be further<br />
refined. An apple is edible of course due to its inherent qualities; however, while<br />
the apple is edible in virtue of its qualities (e.g., freshness), the middle generic<br />
construction does not refer directly to these qualities, but rather, it refers to, if<br />
you like, a derived property of the apple, i.e., being edible because of its qualities.<br />
Thus, the contextual properties of the apple (e.g. freshness, lack of poison, fell on<br />
the ground but the 5-second rule was applied) are implied, while the derived<br />
property is the fact that it is possible to eat the apple, or that there would be no<br />
negative consequences if a human being were to eat the apple. On the other<br />
hand, in a different context, the above construction would have alternative<br />
modal meanings, such as permission or prescription. Having this picture in mind<br />
simplifies and expands the conception of how the property ascription comes<br />
about in middle generic constructions and allows us to further delineate what<br />
we are looking for in the data. Based on the above analysis, I would like to offer<br />
the following definition for the interpretation of the middle:<br />
(24)<br />
The defining characteristic of the middle generic<br />
interpretation is in fact the realization of the modality of<br />
possibility in non-active generic constructions with<br />
regards to the generic agent.<br />
32
I suggest that the middle interpretation is available only when this modality<br />
reading is available in the discourse. This, I propose, is contingent upon: i)<br />
whether a generic agent is interpreted, ii) whether the linguistic context<br />
provides for the modality of possibility to be realized, as opposed to other modal<br />
readings or lack of modality.<br />
3.2. Interpreting the generic agent in Albanian<br />
The non-active suffix -(h)e, which depends on the linguistic context, can be found<br />
in eight strictly defined constructions: (a) passives, (b) reflexives/reciprocals, (c)<br />
middle verb constructions, (d) generic sentences, (e) generic middle sentences,<br />
(f) prescriptive generic sentences, (g) anticausative sentences and, (h)<br />
impersonal passive sentences. Here I will illustrate some of these constructions<br />
with examples from the corpus, which will generate the grounds for the initial<br />
discussion of the role of the non-active suffix:<br />
(25)<br />
a. Passive (F. Lubonja, Përpjekja E-zine)<br />
ky “oktapod” … ushqe-het … nga ekonomia ilegale<br />
this octopus feed- NACT.3SG by economy. DEF illegal<br />
this “octopus” is …fed… by the illegal economy<br />
b. Reciprocal (K. Mehmeti, Kali i Bardhë )<br />
Popujve të Ballkanit u ka ndodhur<br />
People.DAT AGR Balkan.GEN DAT.CL have.3SG happen.PTCP<br />
të vëllazëro-hen e të miqëso-hen...<br />
to fraternize-NACT.3PL and to befriend-NACT.3PL<br />
33
It has occurred with the Balkan people that they were able to be friends and<br />
brothers…<br />
c. Middle verb construction (K. Mehmeti, “Vitet e Urithit”)<br />
e, kur lësho-he-t terri, pahetueshëm<br />
and when drop-NACT.3SG darkness, unsuspected<br />
shëtisin nëpër rrugët e fshatit<br />
stroll.3PL along street.DEF.PL of village<br />
edhe ngjit-en nëpër tavanet e kullave<br />
and lift- NACT.3PL onto roofs of kulla.PL 13<br />
nga i sodisin shtëpiarët-e-fjetur.<br />
from ACC.CL gaze.3PL the-sleeping-family.members<br />
…and when darkness falls, they stroll unsuspected along the streets of the village<br />
and climb onto the roofs of the kulla-s 7 from where they gaze at the sleeping<br />
dwellers...<br />
d. Middle generic construction (Online Albanian Portal 14 )<br />
Karakteristikat kryesore të peshkut të freskët<br />
characteristic.PL main.F of fish AGR fresh<br />
janë: luspat hiq-en relativisht me vështirësi<br />
be.PRS.3SG scale.PL remove-NACT.3PL relatively with difficulty<br />
The characteristics of fresh fish are: scales are relatively hard to remove…<br />
Kemmer (1993) indicates that the more grammaticalized a morpheme is, the<br />
more functions it can take within a language. The non-active imperfective suffix<br />
in Albanian seems to be an example of this claim, given how each of these<br />
examples conveys a different meaning. However, one can make some<br />
13 Kulla refers to a traditional stone construction typical of northern and northeastern Albania.<br />
34
connections in terms of how these constructions are conceptualized. To<br />
elaborate on this point, I draw on the notion of agent-patient distinguishability,<br />
proposed by Kemmer (1993, 2003). In her study of the typology of middle verbs<br />
(as in 23(c)), Kemmer proposes that humans systematically distinguish between<br />
situation types by “means of a difference in grammatical coding” (2003., p. 90).<br />
Situation types are (Kemmer, 1993:7 in Sansò, 2006:238):<br />
(26) “sets of situational or semantic/pragmatic contexts that<br />
are systematically associated with a particular form of<br />
expression. By semantic/pragmatic contexts I do not mean<br />
‘real world contexts’ existing independently of the language<br />
user; situational contexts include ‘real world’ information, but<br />
that information is necessarily filtered through the conceptual<br />
apparatus of the speaker”.<br />
It is argued that the grammatical coding of meanings is instantiated in a host of<br />
interlinked linguistic phenomena. For example, in order to describe a transitive<br />
event, speakers mark the ‘agent’ and the ‘patient’ roles differently because they<br />
conceptualize this event based on the number of participants as well as the<br />
relationship between them. According to Kemmer, in attempting to predict the<br />
way that the number of participants and their relationships are encoded in the<br />
grammar of a language, we can group different situation types. Transitivity<br />
categories, as she points out, can be defined by the degree of agent-patient<br />
distinguishability, where one and two-participant events are at the edge of a<br />
continuum and middle and reflexive are in between the former categories.<br />
Middle verbs accordingly are defined as types of events where agent and patient<br />
are neither pragmatically, nor conceptually distinguishable. However, agent-<br />
14 http://www.lirenti.com/printthread.php?t=4456 Last accessed: 4 September, 2011<br />
35
patient distinguishability is not the only parameter along which we<br />
conceptualize events. A more inclusive parameter, namely the degree of<br />
elaboration of events is proposed, which involves not only the number of<br />
participants, but all other factors which determine the event structure of a<br />
certain type of event.<br />
Another continuum which can be proposed is that of agent specificity: events can<br />
be categorized as being performed by a concrete agent, or by a generic agent. In<br />
the Albanian imperfective non-active paradigm, the agent specificity continuum<br />
can be applied to obtain the correct interpretation in a sentence with a nonactive<br />
imperfective verb. Along the continuum of agent/patient<br />
distinguishability, the passive and the middle generic constructions with<br />
transitive verbs are more similar than, for example, the middle verb formation<br />
and the middle generic constructions in 25(c) and (d) respectively. In both the<br />
passive and middle generic constructions, the agent and the patient are very<br />
clearly separated (following Kemmer 1993). The non-active marker refers to an<br />
agent, while in middle verbs as in 25(c), the action cannot be generic unless<br />
there are other elements that would construe a generic interpretation, and the<br />
agent and the patient are not clearly distinguished. Impersonal passives and<br />
middle generic constructions are also similar in that they denote a state of affairs<br />
where an intransitive action is performed by everyone or can be performed by<br />
anyone. On the other hand, along the agent specificity continuum, passive<br />
constructions as well as reflexive and middle verb constructions differ from<br />
generic middle/impersonal constructions, in that in the former the agent is<br />
specified, i.e., known, while in the latter it is not. Thus, the correct interpretation<br />
of a sentence with a non-active verb relies on the speaker's understanding of<br />
who/what the non-active suffix refers to.<br />
36
In determining whether the generic agent is accessible we need to understand<br />
how the non-active suffix realizes agent defocusing and whether it is feasible to<br />
suggest some type of generalization regarding their interpretation in discourse.<br />
Sansò (2006) investigates four agent defocusing phenomena, including<br />
periphrastic passives, middle verbs, middle generic constructions as well as<br />
impersonal passive constructions. Based on the features and the status of agent,<br />
patient and the event itself (Kemmer’s (1993) situation types), Sansò suggests<br />
that the different agent defocusing strategies employed in the constructions<br />
under investigation rely on the distinguishability and prominence of the agent.<br />
He investigates three situation types: patient-oriented processes, bare<br />
happenings and agentless generic events. Patient-oriented processes are “twoparticipant<br />
event[s] from the point of view of the patient…[where] the agent is<br />
typically identifiable from the context, or even syntactically encoded as an<br />
oblique, but less discourse-central and individuated than the patient” (p. 238).<br />
This type of event is grammatically constructed as a passive form. A bare<br />
happening is defined as a conceptualization where “none of the participants is<br />
focused: as a result, the event too is characterized by low salience” (p. 241). A<br />
bare happening can be an impersonal passive construction such as 19(a) while<br />
agentless generic situations types encapsulate our middle generic construction.<br />
The following table summarizes the prototypical features of the participants in<br />
an event as well as the event itself (Sansò, 2006: 245):<br />
37
Individuation of the<br />
patient<br />
Individuation of the<br />
agent<br />
Reason for defocusing<br />
agent<br />
Patient-oriented<br />
process<br />
Bare happening<br />
+ – –<br />
± – –<br />
Agent is less topical<br />
than the<br />
patient/unimportant<br />
Agent is<br />
irrelevant/unimport<br />
ant<br />
Agentless<br />
generic event<br />
Agent is generic<br />
Mode Realis Realis Irrealis (deontic,<br />
potential)<br />
Aspect Perfective Perfective Imperfective<br />
Contextual salience of<br />
the event<br />
± – –<br />
The agent defocusing strategies for these three situation types can be re-defined<br />
for our purposes from the perspective of agent specificity focusing only on the<br />
non-active imperfective verb marking under survey. Thus, we distinguish<br />
between patient-oriented situations on the one hand, and bare happenings and<br />
generics on the other. The following definitions are given for the status of the<br />
agent in the three situation types, translated into grammatical constructions:<br />
(27) i. Passive Construction: agent is backgrounded but<br />
present and expressed either explicitely (for example<br />
as a prepositional phrase) or inferred from the context<br />
(not present in the sentence but from the context we<br />
know who the agent is)<br />
ii. Reflexive Construction: agent is present and topical<br />
and carries out an action on him/herself<br />
iii. Middle verb formation: agent is present and topical<br />
and carries out an action by him/herself<br />
38
iv. Generic Middle Construction: agent is in the background<br />
and is generic<br />
Following a discussion of the properties of agent defocusing strategies as<br />
evidenced in corpus data, Sansò concludes that, while a tentative hierarchy of<br />
agent defocusing can be proposed for the recoverability of the agent in<br />
discourse, one-to-one correspondences between form and function in agent<br />
defocusing strategies is to be abandoned, especially when one investigates such<br />
phenomenon cross-linguistically. Taking into account the observed data from<br />
the present study however, it seems that proposing such hierarchy could be<br />
viable if we restrict the field of inquiry. Sansò’s treatment focuses on diverse<br />
forms (periphrastic passives, the generic 'they' or 'one' expressions etc.)<br />
therefore making it more difficult to generalize on form-function<br />
correspondences, because the syntactic manifestations of agent-defocusing<br />
strategies put into play a larger number of factors. In this study however, I only<br />
concentrate on one form, namely the imperfective suffix, thereby focusing only<br />
on the different interpretations of the status of the agent and on how the correct<br />
interpretation can be recovered in context.<br />
In addition, Sansò suggests that the distinguishability of the patient/topic is as<br />
important in understanding the situation type as that of the agent. The reason<br />
for this assumption is the fact that he includes in his analysis the classical middle<br />
verbs (the middle Greek diathesis), where the agent and patient are not<br />
distinguishable. However, from the perspective of agent specificity, the strategy<br />
of agent defocusing in middle verbs does not align well with passives or generic<br />
middles, and therefore should not be included in such tentative hierarchy.<br />
Although it can be argued that there is agent defocusing in functional terms in<br />
the middle diathesis, the 'strength' of the agent is hardly compromised. If we<br />
39
ecall example (25c), it is clear that, while the verb 'climb' is a middle verb, the<br />
agentivity of the individuals climbing is not compromised. The genericity of the<br />
patient/topic/clause on the other hand does play a role in the interpretation of<br />
the sentence. It can be argued that when a non-active clause predicates about a<br />
generic subject, the interpretation of the entire clause will be generic and not<br />
temporally bound (Carlson & Pelletier, 1995). However, it does not affect the<br />
status of the agent as being separate from the topic in generic sentences. This<br />
point can be exemplified in the following example:<br />
(28) a. Apples are usually harvested in autumn<br />
b. Apples are usually harvested by farm workers<br />
Therefore, I maintain that the status of the patient is not relevant in order to<br />
understand agent defocusing in our case. In fact, there need not be a patient at all<br />
in the functional sense, given that, in our data we have encountered numerous<br />
examples of generic middle constructions which predicated about a location, or<br />
about an entire clause. In the spirit of Sansò, I suggest that in order to determine<br />
correctly to which entity the Albanian non-active suffix refers to, we can rely on<br />
a prominence hierarchy of agent specificity. In line with the definitions of the<br />
status of the agent in non-active sentences in (27), I propose the following<br />
accessibility hierarchy:<br />
(29)<br />
agent in passives>agent in reflexives/reciprocals/middle verbs >><br />
agent in generics/ middle generics / impersonals<br />
With this accessibility hierarchy I intend to align the strength of agentivity in<br />
terms of the prominence and specificity of the agent. In passives that contain a<br />
40
formally transitive verb, the agent is implicit and the force of the agent is not<br />
diminished by the topicality of the ‘patient’. In reflexive/reciprocal and middle<br />
verbs the agent is still prominent, specific and also topical, but its force is<br />
diminished due to the simultaneous status of the agent as patient or experiencer.<br />
In generic and impersonals, the force of the agent is highly diminished given that<br />
it is ultimately not specific. We can align along a similar specificity hierarchy the<br />
events themselves. In passives and reflexives/reciprocals and middle verbs, the<br />
events are specific unless there is a distinct element of genericity/habituality,<br />
e.g., an adverbial like always. In middle generics and impersonals on the other<br />
hand, the event is not specific unless there is a temporally-binding element. The<br />
following paragraph from our corpus exemplifies this point (from the Albanian<br />
translation of “Alice in Wonderland”):<br />
(30)<br />
Xhelatii mendonte se nuk [mund të prit-ej]g një kokë nëse kjo nuk<br />
ishte e kapur pas një trupi...Mbretij ngulte këmbë se çdo kokë [mund<br />
të prit-et]g...Mbretëresham theksonte se po [të mos zbato-hej]n<br />
urdhëri menjëherë, do t'i vriste [të gjithë]n.<br />
The executioneri's argument was that you [couldn't cut off]g a head<br />
unless there was a body to cut it off from... The Kingj's argument was<br />
that anything that had a head [could be beheaded]g...The Queenm's<br />
argument was, that if something [wasn't done]n about it in less than<br />
no time shem'd have everybodyn executed, all round. (from the<br />
original English version)<br />
As we can see, a non-active form of the verb produces a middle interpretation in<br />
the first two sentences where there is no prominent agent within or outside the<br />
41
clause. It is worth noting here that although the Albanian translation of this<br />
paragraph includes the overt modal mund, a middle interpretation would hold<br />
even in its absence in both cases (Newmark, Hubbard & Prifti, 1982, p. 30). In<br />
the third sentence, the agent in the conditional clause is recovered at the end of<br />
the sentence (everybody/people present there/concrete entities) therefore the<br />
interpretation of the verb is not a middle but a passive. The original paragraph in<br />
English demonstrates how this interpretation is accurate for the Albanian<br />
translation.<br />
In conclusion, middle generic constructions, generic constructions with<br />
transitive verbs as well as impersonal passive constructions differ from passive<br />
and middle verb/reflexive construction from the point of view of agent<br />
specificity. The prominence of the agent in discourse, i.e., the specificity of the<br />
agent, determines the correct interpretation of what the non-active marker -(h)e<br />
refers to. When there is no prominent and specific agent in discourse, the nonactive<br />
marker is interpreted as referring to a generic/unspecified agent. With<br />
regards to the second criterion of the middle generic interpretation, namely,<br />
interpreting a modality of possibility for the generic agent, we turn to our data.<br />
42
Chapter 4<br />
Generic Middle Constructions in Albanian<br />
4.1. The corpus<br />
The data collected for the purpose of investigating the middle construction in<br />
Albanian contains the following books:<br />
Translated into Albanian<br />
- “The Alchemist”, by Paulo Coehlo<br />
- “Alice in Wonderland” by Lewis Carol.<br />
Written in Albanian<br />
- "Thus walks man” by Kim Mehmeti<br />
- “The years of the mole” by Kim Mehmeti<br />
- “The white horse” by Kim Mehmeti<br />
- “Criticism: between art and science” by Natasha Lushaj<br />
- “His signature was his oath” by Natasha Lushaj<br />
- “Her veil of freedom” by Natasha Lushaj<br />
- “An everlasting fire” by Natasha Lushaj<br />
Moreover, the corpus includes excerpts from various Albanian literary works<br />
and political essays, including the authors: Ismail Kadare, Nazmi Rrahmani,<br />
Kasem Trebeshina, Sabri Godo, and Martin Camaj, Natasha Lushaj, Ernest Koliqi,<br />
Ernesto Sabato, Jasunari Kavabata, Fatos Lubonja 15 and Petro Marko.<br />
15The entire work of Fatos Lubonja’s political essays in his blog Përpjekja was originally included in the<br />
corpus and pre-processed, but it is only partly included in the analysis due to time constraints.<br />
43
4.2. Methodology<br />
The texts were scanned using R software, individuating all those instances of<br />
words which ended with one of the following non-active suffixes:<br />
Present<br />
Imperfective<br />
1SG -(h)em -(h)esha<br />
2SG -(h)esh -(h)eshe<br />
3SG -(h)et -(h)ej<br />
1PL -(h)emi -(h)eshim<br />
2PL -(h)eni -(h)eshit<br />
3PL -(h)en -(h)eshin<br />
Table 2: Non-active suffixes in Albanian<br />
Following this stage, the corpus was pre-processed in order to locate those<br />
occurrences which were in fact instances of verbs in the non-active form, instead<br />
of other elements in a sentence with the same ending. Upon having identified<br />
sentences containing non-active verbs, the corpus was further filtered in order<br />
to distinguish generic sentences from other types of non-active sentences, like<br />
those explained in Chapter 3, i.e., middle verbs, passives, reflexives and others.<br />
The next step in the filtering process was the discrimination of different types of<br />
generic sentences such as impersonal generics and middle generic sentences.<br />
The filtering process was based on the judgements of the author of this thesis. It<br />
should be pointed out that often the judgements with regards to these generic<br />
sentences are quite subtle, especially when determining the type of modality in<br />
generic modal sentences. Thus, the methodological downside of this work is very<br />
obvious: only one native speaker (the author of the thesis) has proposed<br />
judgements on the corpus sentences. However, much time was spent on<br />
understanding the text itself so that the judgements on the interpretations can<br />
be justified if we take into account the linguistic context of the phrase.<br />
44
Several difficulties were encountered in the process of determining whether a<br />
middle interpretation arises in sentences with non-active verbs in the filtered<br />
data. As it was explained above, the sentences selected for this analysis, are<br />
either clearly eventive or clearly generic. While it was simple to distinguish<br />
between eventive and generic sentences, it was more problematic to distinguish<br />
between non-active sentences about a certain state of affairs, e.g.,<br />
habitual/generic sentences and non-active sentences with a middle<br />
interpretation, as well as non-active sentences with a different modal<br />
interpretation (e.g., prescriptive). The only viable way in which to distinguish<br />
between these different interpretations was to rely on the possibilities afforded<br />
by context, and it is acknowledged that in various cases the judgements are open<br />
for debate. The following example from our corpus illustrates a case where<br />
interpretation is not straightforward:<br />
(31) [i jati]…vetëm pas tri vitesh vajti ta takonte në burgun e Burrelit,<br />
bashkë me Nuren. Ata të dy kanë ndenjur babë e bir për më shumë<br />
se dy orë e kanë biseduar vetëm. Ç'mund të kenë thënë? Fetiu për<br />
atë kohë sikur ka në kujtesë një fushë të bardhë ku nuk<br />
shkruhet asnjë fjalë. Pas pak muajsh lirohej.<br />
[his father] only after three years went to visit him in the prison<br />
of Burrel, together with Nure. Father and son talked alone for<br />
more than two hours. What might they have said? About that<br />
time, it is as if in Feti’s memory, there is a white field where<br />
nothing is written/can be written. He would be released in a<br />
few months.<br />
45
In the first sentence, the clause ku nuk shkruhet asnjë fjalë can potentially have<br />
the following interpretations:<br />
(32)<br />
i. where nobody has written anything (passive construction)<br />
ii. where it is impossible to write anything (middle generic construction)<br />
iii. where it is not allowed to write anything (prescriptive generic)<br />
iv. where nothing is written<br />
(characterizing generic)<br />
Out of these four possibilities, ii) and iv) are possible interpretations. The other<br />
two are not interpretable due to, a. lack of an expressed discourse agent to which<br />
the non-active suffix can refer to, b. contextual information regarding the<br />
individual in question (i.e., we know the person talked about in this section is a<br />
former political prisoner who is now mentally disabled due to harsh prison<br />
conditions and treatment), which excludes the prescriptive interpretations.<br />
However, the ability and simple generic interpretations are equally viable: the<br />
modality interpretation can be argued to derive as expressing the inability of this<br />
individual to write (i.e., narrate) his sufferings due to shock and trauma, or the<br />
inability of any other person to be able to remind Feti about that time. The<br />
simple generic interpretation can be derived if we understand the locative clause<br />
to further elaborate on the 'resultative' reading, i.e., the present emptiness in the<br />
memory of Feti about those traumatic experiences. However, oftentimes the<br />
interpretation can be quite obvious, as in the following paragraph from our data:<br />
(33)<br />
Para disa kohësh Ben Blushi, duke përshkruar gjendjen në partinë e<br />
tij, përdori një metaforë sipas së cilës PS është si një autobuz ku<br />
shkruhet “mos i fol shoferit”.<br />
46
A while back, Ben Blushi, in describing the situation of his party,<br />
used a metaphor according to which the SP is like a bus where it<br />
is written "do not speak to the driver".<br />
In conclusion, the methodology applied is not strict, given that at such fine level<br />
of analysis and interpretation it is difficult to dissociate sentences from their<br />
context. Contextual/discourse knowledge as well as linguistic metaunderstanding<br />
both play a role in arriving at the right possible interpretation of<br />
the non-active form. The variety and diversity of these factors have made the<br />
analysis of the sentences a rather challenging task. An attempt was made to<br />
select only those phrases where the ability reading is highly unequivocal and in<br />
various cases other native speakers were solicited to interpret the selected<br />
sentences. However, given the subtlety of analysis and the differences in<br />
understanding the context of the texts, it is not excluded that other native<br />
speakers might have different interpretations from mine.<br />
4.3 Results<br />
7,662 occurrences of non-active imperfective verbs were found in the corpus,<br />
out of which 586 sentences were selected. These either (i) produced an<br />
unambiguous modality-of-ability interpretation, or (ii) produced a generic<br />
interpretation, or (iii) were considered interesting examples for comparison<br />
purposes. In other words, the sentences were chosen based on how/whether<br />
they exemplified a typical example of, or typical contrast with, the property<br />
ascription of middles as defined above. Out of the 586 sentences in the corpus,<br />
120 receive an ability interpretation, and 55 do not include adverbial<br />
modification. Below a table is given of verbs that occurred in bare middle generic<br />
47
constructions:<br />
Verbs in bare<br />
middles<br />
Feel 15<br />
Understand 13<br />
Hear 7<br />
See 7<br />
Explain 3<br />
Distinguish 16 (dalloj) 2<br />
Appear 2<br />
Build 2<br />
Distinguish (shquaj) 1<br />
Achieve 1<br />
Notice 1<br />
Crush 17 1<br />
Total sentences N=55<br />
These sentences convey a modality of ability interpretation that is unequivocal.<br />
However, out of the filtered 586 sentences, there were also cases of generic<br />
sentences with the same verbs as the ones above, where the modality of ability is<br />
not available. Consider the following contrast:<br />
(34)<br />
a. Kudo që bënte biseda<br />
Everywhere that<br />
do.IPFV.3SG talk.PL<br />
dëgjohej me vëmendje dhe respekt<br />
hear-NACT.PST.3SG with attention and respect<br />
In any place he held talks he was heard with attention and respect.<br />
16 The verbs dalloj and shquaj are synonymous; dalloj refers to being able to tell something apart from the<br />
surroundings, or to distinguish between entities; shquaj refers to being able to see something with one's<br />
own eyes which is far away or hidden from direct view.<br />
17 The verb thërmoj, 'cause something to break into small pieces’; it does not necessarily involve the<br />
48
. Në fshatin tim... zërat<br />
in village 3SG.POSS voice.PL<br />
dëgjohen<br />
hear-NACT.PRS.3PL<br />
qartë<br />
clear.ADV<br />
nga një lagje në tjetrën<br />
from one neighborhood to next<br />
In my village…voices can be heard clearly from one neighborhood<br />
to the next<br />
In the first example, the interpretation is a passive one with an unspecified<br />
agent: it was the case that the participants in the talks listened to the individual<br />
in question with attention and respect. The backgrounded agent in this sentence<br />
is recovered by the contextual queues that suggest an audience: a talk, as well as<br />
the manner in which he was heard. The same interpretation would hold if the<br />
verb would be in the present tense. In the second example, the agent is not<br />
recoverable from the context and therefore can only be interpreted generally.<br />
The middle interpretation would have been equally viable without the adverbial<br />
qartë.<br />
Cognition and perception verbs ‘feel’, ‘understand’, ‘hear’, ‘see’, ‘explain’,<br />
‘distinguish’, and ‘appear’ (Onishi 2001; Comrie & van den Berg, 2006;<br />
Malchukov, 2008) 18 are the most frequent verbs that produce the ability<br />
physical strength that can be inherent in ‘crush’<br />
18 The verb 'dukem' which is translated 'appear' in this thesis is a special verb in two senses: it is a<br />
deponent verb, i.e., it does not appear in the active form; and its lexical meaning suggests 'ability'. A more<br />
accurate translation would be "able to be visible"<br />
49
interpretation in the non-active constructions where they appear. These verbs<br />
have an ability interpretation even in constructions where they are<br />
complemented by an entire clause. We illustrate this in the following example<br />
with the verb ndiej ‘feel’:<br />
(35)<br />
Ndi-het se pozicioni nuk është më<br />
feel-NACT.3SG that position.DEF not be.3SG more<br />
ai i poh-uesit të çiltër të forcës<br />
3SG.M of assert-NMLZ AGR candid of strength<br />
së<br />
of<br />
njeriut...<br />
man…<br />
It can be felt that the position [of the author] is no longer one of candid<br />
affirmation of the strength of mankind…<br />
A marginal case of ability reading would be the verb kuptoj ‘understand’, which<br />
is ambiguous since it can be interpreted as producing a modality-neutral<br />
impersonal verb or an ability reading. A case in point is the example below:<br />
(36) Kupto-het që në art veprojnë<br />
understand-NACT.3SG that in art act.3PL<br />
disa ligje të tjera…<br />
some law.PL AGR other.PL<br />
50
It is understandable/It is understood that in art, other laws<br />
operate…<br />
In (36), the ability interpretation of the non-active verb is at the margins<br />
between the ability reading and the impersonal reading (in a free translation, the<br />
non-active verb could very well translate in 'it is clear that…').<br />
Of course, not all the verbs in the above list are verbs of cognition. There is also<br />
the verb ndërtoj, ‘build’ and thërmoj ‘smash’. Again, context here plays the crucial<br />
role in achieving the correct interpretation: the ability reading is available due to<br />
the contrastive nature of the coordinate clauses, where 'destroy' and 'build' are<br />
in opposition. Indeed, in uttering this sentence, the second clause will be<br />
pronounced in contrastive intonation:<br />
(37)<br />
Edhe Nuro e Sadik Bylyshit u - a<br />
and Nuro and Sadik Bylyshi.DAT DAT.CL-ACC.CL<br />
prishi shtëpitë (tërmeti)...<br />
destroy.PST.3SG house.PL (earthquake)<br />
POR ndërto-hen PRAPE ato<br />
BUT build-NACT.3SG AGAIN 3SG.F.NOM<br />
“Also Nuro and Sadik Bylyshi’s houses were destroyed [by the earthquake], but<br />
they can be built again.”<br />
With regards to the verb ‘crumble’, it is found in the following context:<br />
51
(38)<br />
…mermeri i palëvishëm është historia me të dyja kuptimet: edhe fakti,<br />
ngjarja që nuk zhbëhet më, edhe historia e shkruar, që edhe ajo është<br />
shumë vështirë të ribëhet...ky raport vjen...duke u përmbysur ...madje<br />
në fund, më lehtë mund të besosh se thërmohet mermeri...<br />
...the immovable marble represents history in both senses: the fact,<br />
the event that can't be undone, and also the written history, which is<br />
also very difficult to remake...this rapport is<br />
progressively...reversed...in fact, at the end, it will be easiler to believe<br />
that the marble can crumble 19<br />
Here again the ability reading can be interpreted as occurring throughout the<br />
paragraph as exemplified in the preceding non-active verbs, and can therefore<br />
be easily interpreted in the case of the verb 'crumble'.<br />
At the same time, not all verbs with an inherent experiencer agent can obtain an<br />
ability reading. For example, there is no instance in the corpus of the verb 'know'<br />
giving rise to the ability reading, although of course a 'generic' agent reading, or<br />
an 'impersonal reading' is possible, as can be seen in the (randomly selected)<br />
sentence below:<br />
19 This paragraph is taken from N. Lushaj's commentary on a poem by I. Kadare. In order to better explain<br />
the context, it must be noted that throughout this section, Lushaj describes the difficulty with which truth<br />
about Laocoön, the subject of the poem, is unfolded verse after verse. Therefore, at the end of the paragraph<br />
the author concludes that it seems easier to conceive that a marble statue of Laocoön could crumble and he<br />
could appear to us to recount his story in person, than to conceive that truth can emerge unscathed by the<br />
manipulations of history. Unfortunately this poem does not seem to be translated into any foreign language,<br />
but can be found in Kadare (1976) Selected Works. Shtypshkronja e Re, Tiranë.<br />
52
(39) Di-het se me anë<br />
know-NACT.3SG that through<br />
të saj [poezisë]<br />
AGR 3SG.F [poetry]<br />
njeriu pushtoi botën për herë të parë...<br />
man conquer.PST.3SG world for time AGR first<br />
“It is known that through it [poetry] man conquered the world for<br />
the first time....”<br />
In the corpus, some verbs were individuated which, it seems, can’t be in a middle<br />
construction, not only in a non-modified middle, but even in a modified one.<br />
These verbs are given in the following table:<br />
Verbs which can’t be in<br />
Definition<br />
middles 20<br />
Ndodhet<br />
to be located somewhere<br />
lartohet to stand erect 21<br />
Përpiqet<br />
to try<br />
Përbëhet<br />
to be composed of<br />
zhvillohet<br />
to occur<br />
Gdhihet<br />
to become dawn<br />
ngryset<br />
to become dusk<br />
Nxitohet<br />
to hurry<br />
Shpeshtohet<br />
to happen more often<br />
Dihet<br />
to come to be known<br />
Quhet<br />
to be named<br />
20 All the verbs are presented in their 3rd person non active form in the present tense.<br />
21 Usually about an object in nature such as a tree<br />
53
All the verbs which can't be in a middle construction are stative verbs, however,<br />
they have another aspect in common, in that the majority of them have an<br />
unaccusative/impersonal meaning, i.e., there is no need for an inherent agent to<br />
perform the action.<br />
54
4.4. Analysis and discussion<br />
Fellbaum (1985) notes that in English, verbs of perception, doubt, emotion, etc.<br />
can’t form a middle construction. Such claim is also supported by Fagan (1992)<br />
about English and German. Condoravdi (1989) on the other hand points out that<br />
if examined from a cross-linguistic perspective, the middle is a notion, a specific<br />
meaning, and it is not viable to attempt to define a class of verbs that are<br />
available in this construction, especially due to the pragmatic constraints that<br />
are very much affecting the successful realization of the construction. I support<br />
Condoravdi's argument, especially in light of the data presented above: the more<br />
I investigated the possible interpretations of different verbs that were<br />
encountered through the filtering process, the more I realized that given a<br />
specific context, a majority of these verbs would be able to form a middle<br />
construction in Albanian, especially if they were modified middles, containing an<br />
adverb or negation. It was for this precise reason that I decided to narrow down<br />
the types of middles under investigation and only focus on the non-modified<br />
ones. This latter investigation seems to have produced some interesting results,<br />
which were truly unexpected when I began to work on this topic.<br />
Firstly, I believe that a much clearer and simpler definition of the middle<br />
interpretation was proposed: where the realization of the modality of ability<br />
with regards to the implied generic agent is the demarcation line between<br />
generic constructions or constructions of different kinds of modality and the<br />
middle generic construction. This definition seems to be borne out from the data<br />
since the predicates which don't allow or don't necessarily need a [+human]<br />
agent in their argument structure are not available for a middle construction in<br />
Albanian.<br />
Secondly from our data it was clear that the presence of the generic agent in the<br />
predicate's argument structure is not the only condition for the realization of the<br />
55
modality of ability interpretation. As was shown, the verb 'know' is not available<br />
for this construction, even though it requires the presence of a [+human]<br />
argument. What the verbs that were shown to be able to realize this construction<br />
in our data have in common is also a type of unboundedness of the event<br />
described by the psychological verbs. Apart from an experiencer agent, verbs<br />
like ‘feel’, ‘understand’, ‘hear’, ‘see’, ‘distinguish’ etc, all involve some kind of<br />
process in their meanings, which an experiencer agent goes through. The verb<br />
‘know’ on the other hand, has a punctual meaning, as the boundary between<br />
knowing and not knowing can be termed conceptually very clear. Similarly, the<br />
verb 'be named' is grammaticalized as a type of copular verb and in that sense is<br />
a stative one, i.e., it does not include in its meaning the process by which<br />
someone is named.<br />
A functional-cognitive approach to the middle realization (Davidse & Heyvaert,<br />
2007) proposes that the meaning of the middle is derived from a process of<br />
subjectivization of an "agentive-patientive process-participant relation into an<br />
activo-passive letting modality” (p. 40). In other words, contexts that allow for<br />
the realization of a middle construction contain elements such as a protopatientive<br />
element (which can also be a locative in their view) as well as a<br />
generic agent (which does not have to be a human agent). This context allows<br />
speakers to conceptualize an event involving such entities into an active-passive<br />
formation, where the proto-patientive element realizes the ‘letting’ modality<br />
(Talmy 2000), i.e., it is conducive to the realization of the action. What this seems<br />
to suggest is that the grammatical subject of the middle is the element which is<br />
crucial in the realization of a middle construction. Crucially, in this work, the<br />
orientation of the modality involved in a middle construction is NOT assigned to<br />
the generic agent. As such, the level of clausal organization in a middle<br />
construction is no longer a predicative one (i.e., statements about states of<br />
affairs) but an interpersonal one (Halliday 1985; Hengeveld 1989; McGregor<br />
56
1997). At the interpersonal level of discourse, individuals rely on shared notions<br />
of cause and effect and other such basic elements, in order to formulate some<br />
type of metaphorical description of the relationship between the grammatical<br />
subject and the verb in a middle generic construction, which is then<br />
grammatically encoded as an actor-patient relationship. They provide the<br />
following example to elaborate on this point (p.71):<br />
(40) That organic flour bakes delicious bread.<br />
While the above sentence of course characterizes a relationship between the<br />
properties of the flour and the quality of the bread, the verb implies an agent<br />
through its reference to an instrument.<br />
This treatment provides, in my opinion, a crucial insight about the functional<br />
mechanisms that allow for the middle interpretation to be realized, namely, the<br />
identification of the interpersonal level of clause organization as the most<br />
appropriate one in which the middle should be treated. In Albanian however,<br />
based on the data, what I have found is that, given the necessity for an implied<br />
generic agent for the realization of the modality of ability that is crucial for the<br />
middle realization, it is more viable to characterize the subjectivization of such a<br />
process-participant situation as the subjectivization of the experiencer generic<br />
agent implicit in middles. In addition, our data shows that there are some<br />
restrictions in terms of the semantics of the predicates which allow or never<br />
allow for a middle interpretation to arise. Our data indicates that agentexperiencer<br />
verbs do not all realize the middle interpretation in a non-modified<br />
setting, but differ with regards to the notion of process: if these verbs denote a<br />
process, they tend to allow for the modality of ability to be interpretable.<br />
The discourse dependency of the ability interpretation is another interesting<br />
issue which emerged during the study of the data. What is clear is that the<br />
57
modality of ability that could be interpretable in different verbs (not necessarily<br />
psychological verbs as above) depends also on context boundaries: if any<br />
element in the linguistic context restricts the unboundedness of the generic<br />
agent or of the process described by the verb, then the ability interpretation of<br />
non-active imperfective verbs in Albanian is not available.<br />
4.5. Conclusions<br />
Analysing the middle construction in this Albanian corpus presented some<br />
difficulties. Firstly, starting from absolutely raw data gathered online or through<br />
contacts with the authors, it was a lengthy process to have the data analyzable in<br />
an acceptable format. However, the development of this corpus was not<br />
sufficiently elaborate to allow for tagging of syntactic functions in specific<br />
sentences and therefore, I had to rely only on the verbal suffix as the only way to<br />
scan through the data and identify relevant constructions. Secondly, given that<br />
middle constructions are dependent on the non-active verbal paradigm of<br />
Albanian, during the pre-processing stage, a great deal of time and effort was<br />
given to discarding verbs from non-verbs, and most importantly, to choosing<br />
relevant examples of non-active verbs. Since this project had an exploratory<br />
nature and did not start with pre-assumptions about middles in Albanian or<br />
about the types of verbs that could be involved in it, the structural flexibility of<br />
Albanian sentences was seen as an asset rather than a liability. However, the<br />
richness of structure manifested in the corpus it was very difficult to provide<br />
coherent judgements from untagged data.<br />
Secondly, a significant problem was posed by the undefined nature of the middle<br />
construction in terms of interpretation. Examples given in previous literature<br />
covered different types of generic constructions and most of the examples were<br />
sentences provided by researchers themselves, rather than a corpus. Therefore,<br />
58
finding comparable constructions in a corpus of natural language was from the<br />
start improbable, or at least very difficult. Indeed the definition of middles as<br />
having the crucial property of modality was arrived at only after a thorough and<br />
lengthy review of the data itself.<br />
Finally, the project did not involve any kind of direct judgements from other<br />
native speakers of Albanian, either to test the interpretations of the<br />
psychological verbs interpreted as bearing a modality interpretation above, or to<br />
test other types of verbs that could potentially obtain an ability reading provided<br />
they were in an adequate linguistic context.<br />
59
Chapter 5<br />
General Conclusions and future research<br />
The present project presents an overview of the different verbs and contexts<br />
which allow for a middle interpretation in Albanian. Firstly, the presence of the<br />
generic agent in a middle construction is crucial for the realization of its<br />
meaning, and the modality that is interpreted in these bare middle generic<br />
constructions is that of an ability of the generic agent. The generic agent in<br />
discourse can be identified based on a hierarchy of agent specificity: the nonactive<br />
imperfective verb will refer always to the most prominent and specific<br />
discourse agent. In cases where a specific agent lacks in discourse, the reference<br />
can only be interpreted as generic. Verbs which don’t specify a human agent in<br />
their argument structure are not available for such interpretation. Secondly, it<br />
turns out that a specific type of agent is prone to realize the middle<br />
interpretation more than other types, namely, an agent with an experiencer<br />
semantic role. Thirdly, the middle interpretation is sensitive to boundedness:<br />
given that it is a generic sentence it is of course clear that telic situations are in<br />
contrast to its meaning. However, even in a non-telic situation, non-active<br />
imperfective verbs do not always provide an ability reading. This last issue is<br />
somewhat unclear, and the corpus was not explored at sufficient depth in order<br />
to resolve this ambiguity.<br />
In conclusion, based on the findings of this study, it seems that middle<br />
constructions as can be found in natural language discourse in Albanian can be<br />
considered an extension of the interpretation available in bare middle generic<br />
constructions with experiencer agent verbs. The subjectivization of the generic<br />
agent allows for this interpretation to be carried on by different verbs as long as<br />
60
the discourse does not provide/include event-delimiting elements, and as long<br />
as other modality alternatives are excluded by the context.<br />
Future research on the middle generic construction in Albanian, and potentially<br />
in other syntactically similar languages, I believe should be on further defining<br />
the definition of the middle construction based on natural language, and mainly<br />
spoken language. Moreover, it would be very useful if the present corpus were<br />
further utilized. For this purpose an effort in tagging the data would facilitate<br />
research and would provide us with more definite conclusions about the<br />
difference between generic sentences that bear modality vs. those which don't.,<br />
as well as to test the hypothesis of agent specificity hierarchy presented in this<br />
thesis.<br />
61
Bibliography<br />
Ackema, P. & Schoorlemmer, M., 2005. Middles. In: Everaert, M., Van Riemsdijk,<br />
H., Goedemans, R., & Hollebrandse, B. (eds). The Blackwell Companion to<br />
Syntax, Volume III, Blackwell Publishing<br />
Ackema, P. & Schoorlemmer, M., 1994, The middle construction and the syntaxsemantics<br />
interface. Lingua, 93, pp. 59-90. North-Holland<br />
Camaj, M. 1984. Albanian Grammar: with exercises, chrestomathy and glossaries.<br />
Harrassowitz, Wiesbaden.<br />
Carlson, G. N., & Pelletier, F.J. (eds.), 1995. The Generic Book. University of<br />
Chicago Press<br />
Chomsky, N. and H. Lasnik. 1993. The theory of principles and parameters. In J.<br />
Jacobs et al. (eds.) Syntax: An International Handbook of Contemporary<br />
Research, Vol. 1. Walter de Gruyter, pp. 506-569.<br />
Comrie, B. & van den Berg, H., 2006. ‘Experiencer constructions in Daghestanian<br />
languages’. In Ina Bornkessel, Matthias Schlesewsky, Bernard Comrie,<br />
Angela D. Friederici (eds.): Semantic Role Universals and Argument<br />
Linking: Theoretical, Typological, and Psycholinguistic Perspectives, 127–<br />
154. Berlin: Mouton de Gruyter.<br />
Condoravdi, C., 1989. The middle: where semantics and morphology meet. MIT<br />
Working Papers in Linguistics 11:18–30.<br />
Davidse, K., Heyvaert, L. 2007. On the middle voice: an interpersonal analysis of<br />
the English middle. Linguistics, 45 (1), 37-82.<br />
Fagan, Sarah M. 1992.The syntax and semantics of middle constructions: A study<br />
with special reference to German. Cambridge: Cambridge University Press.<br />
Fellbaum, C. (1985). On the middle construction in English. Indiana University<br />
Linguistics Club.<br />
Fellbaum, C. and Zribi-Hertz, A. 1987. The Middle Construction in French and<br />
English: A comparative Study of its Syntax and Semantics, manuscript,<br />
Princeton University and Université de Paris 8.<br />
62
Halliday, M.A.K. 1985. An Introduction to Functional Grammar. London: Arnold.<br />
Hatcher, A.G.,1943. Mr. Howard Amuses Easy, Modern Languages Notes Vol. 58.<br />
No. 1, p., 8-17.<br />
Haspelmath, M. & Müller-Bardey. 2004. Valency Change. In: Booij, G. & Lehmann,<br />
C. & Mugdan, J., HSK-Morphology: A Handbook on Inflection and Word<br />
Formation<br />
Heim, I. 1982. The Semantics of Definite and Indefinite Noun Phrases, University<br />
of Massachusetts: Ph.D. dissertation; published 1989, New York: Garland.<br />
Available at http://semanticsarchive.net/Archive/Tk0ZmYyY/ , last<br />
accessed 28 August, 2011<br />
Hengeveld, K.1989. Layers and operators in Functional Grammar. Journal of<br />
Linguistics, 25, 127–157.<br />
Hoekstra, Teun and Ian Roberts. 1993. Middle Constructions in Dutch and<br />
English. In Knowledge and Language. Vol. 2: Lexical and Conceptual<br />
Structure, ed. Eric Reuland and Werner Abraham, 183-220, Dordrecht:<br />
Kluwer<br />
Jespersen, O. 1927. A Modern English Grammar on Historical Principles.<br />
Heidelberg, C. Winter<br />
Kemmer, S. 2003. "Human Cognition and the Elaboration of Events: Some<br />
Universal Conceptual Categories." The New Psychology of Language, Vol. 2.<br />
Kemmer, S 1993. The Middle Voice. Amsterdam: John Benjamins Publishing<br />
Company<br />
Keyser, S. J. & Roeper. T. 1984. On the Middle and Ergative Constructions in<br />
English. Linguistic Inquiry 15:381-416.<br />
Manfred Krifka. An Outline of Genericity. SNS-Bericht, 87-25, 1987<br />
Lekakou, M. 2005., In the Middle, Somewhat Elevated: The semantics of middles<br />
and its crosslinguistic realization. PhD, University College London, UK.<br />
Available at LingBUZZ,<br />
http://ling.auf.net/lingbuzz/@EtLdRQGGAFZGOZqW/LZdkwsyy?12, last<br />
63
viewed on 21 October 2010<br />
Lekakou, M. 2002. Middle semantics and its realization in English and Greek.<br />
UCLWPL, Vol: 14<br />
Lewis, M. P. (ed.), 2009. Ethnologue: Languages of the World, Sixteenth edition.<br />
Dallas, Tex.: SIL International. Online<br />
version: http://www.ethnologue.com/.<br />
Levin, B. and M. Rappaport Hovav (2011) “Lexical Conceptual Structure”, in K.<br />
von Heusinger, C. Maienborn, and P. Portner, eds., Semantics: An<br />
International Handbook of Natural Language Meaning, Vol. I. Mouton de<br />
Gruyter, Berlin.<br />
Malchukov, A. 2008. Split intransitives, experiencer objects and<br />
‘transimpersonal’ constructions: (re-) establishing the connection. In. M.<br />
Donohue & S. Wichmann (eds). Typology of languages with semantic<br />
alignment. Oxford University Press. 76-101.<br />
McGregor, W.B. 1997. Semiotic Grammar. Oxford: Clarendon.<br />
Newmark, L., et.al. 1982., Standard Albanian: a reference gramar for students.<br />
Standford University Press<br />
O’Grady, W. 1980. The derived intransitive construction in English. Lingua, 52, p.<br />
57-72<br />
Perlmutter, D., 1978. Impersonal passives and the Unaccusative Hypothesis. In: J.<br />
Jaeger et al. (eds.), Proceedings of the fourth annual meeting of the<br />
Berkeley Linguistics Society, 157-189.<br />
Rapoport, T. R., 1999. The English Middles and Agentivity. Linguistic Inquiry, Vol.<br />
30, No. , p. 147-155<br />
Sansò, A. 2006., Agent defocusing revisited: Passive and Impersonal<br />
Constructions in some European Languages. In: Abraham, W. & Leisiö, L.,<br />
Passivization and Typology: Form and Function, Amsterdam: Benjamins,<br />
239-270.<br />
Stalmaszczyk, P. 1993. The English Middle Construction and Lexical Semantics.<br />
64
Papers and Studies in Contrastive Linguistics Vol. XXVII, Available at<br />
http://ifa.amu.edu.pl/psicl/files/27/09Stelmaszczyk.pdf, last accessed<br />
28 August, 2011<br />
Schubert, L.K. & Pelletier, F.J., 1989. "Generically speaking, or, using discourse<br />
representation theory to interpret generics", in G. Chierchia, B. Partee, R.<br />
Turner (eds.), Property Theory, Type Theory, and Semantics, Vol. II,<br />
Kluwer, 193-268<br />
Talmy, L. 2000. Towards a Cognitive Semantics. Vol. 1. MIT Press<br />
Vendler, Z. 1967. Linguistics in philosophy. Ithaca, New York: Cornell University<br />
Press<br />
65
Appendix: Generic middle constructions from corpus<br />
1. Kuptohet, do ishte e padrejtë dhe mëkat...<br />
2. …pasqyrë ku do të shihej fytyra e një demokracie të deformuar<br />
3. thuajse askund nuk dëgjohej zëri i atyre të cilët do e zbusnin acarimin e<br />
marrëdhënieve ndëretnike<br />
4. shihej qartë se të gjithë mbronin vetëm etnitetin e vet...<br />
5. ...nga çdokundë dëgjohej ulërima e çmendurisë shoviniste serbe<br />
6. vitet e fundit të dekadës së tetë të shekullit të njëzet, ishin pasqyrë ku<br />
qartë shiheshin fytyrat e moshuara të nacionalizmave të shumtë...<br />
7. sa herë që VMRO-DPMNE-ja sundonte, PDSH-ja ishte "ligj" i vetëm për<br />
shqiptarët, dhe në Maqedoni ndihej një "dyqeverisje"...<br />
8. Si shpjegohet që flet spanjisht?<br />
9. ke ndonjë mendim si shkohet atje?<br />
10. Çaji, për mendimin e tij, ishte gjithnjë më i mirë kur shërbehej në enë<br />
kristali, sepse kundërmimi ruhej më mirë<br />
11. E kuptoi se rrethana mund të kuptohej edhe në këtë mënyrë...<br />
12. Por karvani u nis për udhë dhe ishte e pamundur të dëgjohej se çfarë<br />
thoshte...<br />
13. Dëgjoheshin klithma, të qara fëmijësh, hingëllima kafshësh...<br />
14. Nga toka nuk nxirrej më gjë...<br />
15. Kur futesh në shkretëtirë nuk kthehesh dot mbrapsht...<br />
16. Devexhiu deshi të dinte se në ç'rrethana të jashtëzakonshme Zoti lejonte<br />
të shihej e ardhmja<br />
17. Rrinte mbi humnerën e thiktë poshtë, duke zotëruar pamjen e gjerë deri<br />
përtej luginës ku shquheshin fshatra të tjerë<br />
18. Nga qendra e fshatit shihej pamja e saj ballore<br />
66
19. Halimi e dinte se kështu i kishte dhënë frymë një zjarri që nuk do të<br />
shuhej lehtë<br />
20. Apo (është) një gjë që s’kuptohet dot në kohët e sotme, kur më lehtë<br />
ndiqet ajo rruga "e lehtë" për të siguruar pak komfort?<br />
21. (jeta e tij) është shkruar jo me fjalë, po me vepra që nuk shuhen kurrë<br />
22. ...në personin e tij dalloheshin tiparet e një djaloshi patriot<br />
23. Pranvera ndihej në erën e re të malit<br />
24. LANÇ në Shqipëri nuk mund të shuhej lehtë<br />
25. Edhe tani dallohen në fund të faqes së Qores...ca brezare me gurë...<br />
26. Thonë se në errësirë përreth zjarrit dëgjohen zëra të çuditshëm që<br />
s'kuptohet nga vijnë<br />
27. Në fshatin tim...zërat dëgjohen qartë nga një lagje në tjetrën<br />
28. ...pylli mban mbi vete plagët e një shpate të madhe, që nuk mbyllen kurrë<br />
29. Edhe Nuro e Sadik Bylyshit ua prishi shtëpitë (tërmeti), por ndërtohen<br />
prapë ato.<br />
30. Ndihej era e mishit të djegur...<br />
31. Në qetësinë e asaj nate dëgjoheshin vetëm bisedat me zë të ulët të grave...<br />
32. Por ndoshta ajo ishte e lumtur, megjithëse plagët e shpirtit s’mund t’i<br />
shëroheshin lehtë...<br />
33. Në errësirë nuk dallohet se ku janë vajzat e ku janë djemtë...<br />
34. duken edhe pirgjet e dheut mbi zgjatimet e llogoreve anash që lidhen me<br />
të...<br />
35. disa të tjerë nuk kishin besim se mund të fitohej kundër një armiku kaq të<br />
fuqishëm...<br />
36. E ardhmja e lirë ende nuk kishte ardhur por ndihej nëpër erën e barutit...<br />
37. me derte që nuk harrohen...<br />
38. ato plagë që s’harrohen<br />
39. kjo lidhje mund të vihej re më thjesht në tematikën e veprës së tij<br />
67
40. në përgjethësi ndjehet thyerja që ka pësuar poeti i ri<br />
41. tejembanë ndjehet dhimbja e ndarjes si me thikë<br />
42. vargu është i shtruar, ka një trishtim të madh që nuk vihej re më parë<br />
43. këtij romani do të duhej kohë gjersa t’i gjendej ana, të kuptohej plotësisht<br />
44. janë ngritur mbi shtratin e eposeve e miteve, me...filozofinë epokale të të<br />
cilave shpjegohen shumë gjëra<br />
45. kjo shpjegohet me përpjekjet për të kuptuar shenjat e fshehta të botës<br />
46. kush nuk blihej, kalbej<br />
47. për Kadarenë nuk mund të thuhet asnjëherë fjala e fundit<br />
48. Ato nuk përmblidhen dot në një frazë të vetme<br />
49. Mosbesimi i portave dukej hapur...<br />
50. 'bjeri, bjeri qenit!' - dëgjoheshin nga të katër anët<br />
51. Me ç’mjete (letrare) kalohet nga gjendja e pritjes në atë të zhgënjimit dhe<br />
trishtimit?<br />
52. kryesisht kjo situatë ndërtohet me dialogje...<br />
53. ngjarja nuk zhbëhet më...historia...është shumë vështirë të ribëhet<br />
54. madje më lehtë mund të besosh se thërmohet mermeri...<br />
55. Por, në planin kohor ky raport interpretohet ndryshe<br />
56. ngjarjet...nuk mund të zhbëhen më<br />
57. pa këtë shqetësim poetik e qytetar nuk mund të kuptohet biografia e tij<br />
(Agolli) poetike...<br />
58. ai nuk e ka qejf pozicionin e poetit si një nga zotat e universit, nuk<br />
përpiqet ta hutojë lexuesin me truke e figuracion që, sa më shumë të mos<br />
kuptohet, aq më të mistifikuar bëhet imazhi...edhe [i] vetë krijuesit...<br />
59. ...mund të themi se nuk besohet kollaj se ky pelegrin i thinjur nuk do të<br />
niset përsëri në udhë...<br />
60. ndihet se pozicioni nuk është më ai i pohuesit të çiltër të forcës së njeriut<br />
61. ...ende ndihet lodhja, mërzia dhe vetmia...<br />
68
62. ...motivet e tij ndihen anë e mbanë poezisë së Agollit<br />
63. ...autori nuk e manipulon figurën, ashtu si edhe fshatari që nuk rrotullohet<br />
kollaj...<br />
64. Siç shihet edhe nga sasia e madhe e koncepteve...<br />
65. por në qënien e tij ndiheshin një kënaqësi e madhe dhe një lumturi e<br />
hapur<br />
66. ndihet në kujdesin e autorit për t’i përdorur ato e për të krijuar të tjera<br />
me brumin e lashtë...ajo dhimbje tragjike<br />
67. kuptohet që në art veprojnë ligje të tjera, ndryshe ai do të ishte kopje e<br />
keqe e realitetit...<br />
68. një gjëje që ecën e s'ndalet dot<br />
69. por s'kuptohet ç’fituan prej saj njerëzit<br />
70. ...yll në hapësirën e poezisë e prozës shqipe, e që s'mund të shuhet shpejt<br />
71. ky pozicion i ka dhënë një këndvështrim përtej të zakonshmes, që mund<br />
të fitohet vetëm me sytë e atij që ka vajtur...atje ku askush tjetër nuk ka<br />
guxuar...<br />
72. kuptohet që pikëçuditsja këtu është plotësisht me vend...<br />
73. ja si shpjegohet largimi nga të tjerët, moskuptimi me ta, dhe mospranimi i<br />
realitetit nga heroi romantik<br />
74. ...dhe fshihen më vështirë nga kujtesa<br />
75. kuptohet se objekti i mallit...vjen në mendje në formë kujtimesh...<br />
76. ...(bora) zbret pa zhurmë, nuk dëgjohet...<br />
77. ...rimat fundore....japin pikën kulminante të dhimbjes, që duhet të jetë e<br />
tillë, përndryshe nuk shkohet dot tek akti i flijimit...<br />
78. ...në formën e tyre është ndërtuar konstrukti i brendshëm i shpirtit të tij,<br />
gjë që nuk mund të prishet më kurrë<br />
79. Kështu shpjegohet vdes me gaz...<br />
80. Siç shihet, nuk llogariten si rrokje m[ vete dy zanore kur takohen në fund<br />
69
e në fillim...<br />
81. ...do të thotë se jeta kuptohet, mësohet sesi të jetohet duke u zhytur në<br />
(impersonal), nga eksperienca personale...<br />
82. Tani që u bë plak guxoi, tani ka përvojën e duhur, nga ato që fitohen<br />
vetëm pas një jete të tërë<br />
83. lakmitar edhe për liri, që s'mund të kuptohet e të arrihet pa mirëqënie,<br />
drejtësi e dinjitet<br />
84. Por është një e errët e shndritshme, si ajo e hapësirës ndërujore, ku në<br />
vështrim të parë ndihet kaosi<br />
85. Është një fillim premtues sepse ndjehet talenti origjinal...<br />
86. Në rradhën e metaforave shpesh ndihet motivi i zhgënjimit<br />
87. Ndjehet brenda kësaj shembjeje të përgjithshme edhe një shembje tjetër<br />
88. Kështu kuptohet fëmijëria me heshtjen ndjellakeqe të vetvetes dhe që<br />
s'arrin ta mbysë as era e mandarinës<br />
89. Kuptohet se kur të gjendem në anën tjetër të tokës<br />
90. ...në atë strofkull të ngushtë miu nuk futej as koka e saj<br />
91. ...dera hapej nga brenda...<br />
92. ...por nuk arriti të dallonte asgjë që mund të haej ose të pihej...<br />
93. ...kuptohej se nuk ishte ndonjë dëshirë të merrej as me Lizën dhe as me të<br />
tjerë...<br />
94. ...Midis tyre Liza dalloi edhe Lepurin e Bardhë, i cili fliste me nxitim e<br />
nervozizëm dhe kuptohej se ishte i shqetësuar sepse...<br />
95. Në qendër të Sallës së Gjyqit shihej një tavolinë<br />
96. Ata kishin një thes të madh, që mbyllej nga njëra anë me lidhëse<br />
97. ...në barakën aty pranë dëgjohej tringëllima e filxhanëve të çajit të<br />
Lepurushes Marzolina<br />
98. s'mund të përfytyrohet njeriu pa hije<br />
99. degjoheshin thirrjet e njerëzve të kapur nga ethja e shkatërrimit të<br />
70
pushtuesit<br />
100. ...kështu është njësoj sikur të mendosh se bulimia mund të shërohet duke<br />
ngrënë...ajo në fakt shërohet me të tjera metoda...<br />
101. ...E ky (dinjitet dhe respekt) nuk mund të arrihet pa kultivimin e virtytit<br />
politik për të cilin flet Montesquieu<br />
102. ...paaftësia për të parë e përballuar realitetin e zi që na rrethon nuk mund<br />
të kuptohet plotësisht pa marrë në shqyrtim kulturën që kemi trashëguar<br />
prej komunizmit...<br />
103. ...vetëm kështu mund të shiten gazetat në krizë<br />
104. ...çështja e vetëvrasjes së adoleshentëve nuk është ndonjë gjë e panjohur<br />
që mund të shpjegohet vetëm me magjira dhe me punët e të paudhit...<br />
105. ...kuptohet se pastaj më vërshuan nëpër tru pluhurat që hamë, ndotja<br />
akustike, ndotja vizuale...<br />
106. Rritja e masave për të penguar hyrësit kuptohet se edhe ne na largon...<br />
107. Lehtësia me të cilën trajtohet ky problem vihet re më së pari në mënyrën<br />
se si analizohet humbja<br />
108. Për fat të keq edhe nga LSI, që luftën kundër korrupsionit e kishte një nga<br />
pikat bazë të programit të saj, nuk dëgjohet ndonjë zë në këtë aspekt<br />
109. në Shqipëri është krijuar tashmë tradita se profesioni që mund të arrihet<br />
më shpejt pa qenë nevoja të bëhesh diçka duke investuar intelektualisht<br />
dhe moralisht është ai i politikanit<br />
110. Shtoi kësaj edhe traditën e trajtimit të elektoratit si një tufë debilësh që<br />
manipulohen lehtë<br />
111. sipas meje kjo masë kritike do të arrihej kur shumica e kësaj kategorie do<br />
të humbnin besimin se qeveria po punon për ta sikurse ka premtuar<br />
112. njerëzit nuk janë budallenj që mund të mashtrohen lehtë<br />
113. Ato (mediat) duhet të kenë një identitet të tyrin që krijohet nga mbledhja<br />
së bashku e një ekipi gazetarësh të cilët ndajnë afërsisht të njëjtat<br />
71
aspirata...<br />
114. Si shpjegohet kjo?<br />
115. Asnjë makinë nuk riparohet duke debatuar për të nga lulishtja përballë<br />
116. Dhe po të jetë i qartë ky vullnet, zgjidhet shumë më lehtë edhe problemi i<br />
mënyrës së shkarkimit të Prokurorit...<br />
117. E kuptohet se kjo kërkon kohë<br />
118. ky do të ishte fillimi i një krize ekonomike dhe financiare, që, gjithë sipas<br />
guvernatorit, ndoshta do të krahasohej me atë të fundit që ka kaluar<br />
Shqipëria<br />
119. (njohja intuitive) nuk mund të fitohet veçse duke e jetuar përditë një<br />
realitet<br />
120. ...mungojnë dy gjëra kryesore që të arrihen konsensuse midis<br />
kundërshtarëve<br />
72